diff --git a/src/backend/app/algorithms/evaluates/clip/__init__.py b/src/backend/app/algorithms/evaluate/clip/__init__.py similarity index 100% rename from src/backend/app/algorithms/evaluates/clip/__init__.py rename to src/backend/app/algorithms/evaluate/clip/__init__.py diff --git a/src/backend/app/algorithms/evaluates/clip/bpe_simple_vocab_16e6.txt.gz b/src/backend/app/algorithms/evaluate/clip/bpe_simple_vocab_16e6.txt.gz similarity index 100% rename from src/backend/app/algorithms/evaluates/clip/bpe_simple_vocab_16e6.txt.gz rename to src/backend/app/algorithms/evaluate/clip/bpe_simple_vocab_16e6.txt.gz diff --git a/src/backend/app/algorithms/evaluates/clip/clip.py b/src/backend/app/algorithms/evaluate/clip/clip.py similarity index 100% rename from src/backend/app/algorithms/evaluates/clip/clip.py rename to src/backend/app/algorithms/evaluate/clip/clip.py diff --git a/src/backend/app/algorithms/evaluates/clip/model.py b/src/backend/app/algorithms/evaluate/clip/model.py similarity index 100% rename from src/backend/app/algorithms/evaluates/clip/model.py rename to src/backend/app/algorithms/evaluate/clip/model.py diff --git a/src/backend/app/algorithms/evaluates/clip/simple_tokenizer.py b/src/backend/app/algorithms/evaluate/clip/simple_tokenizer.py similarity index 100% rename from src/backend/app/algorithms/evaluates/clip/simple_tokenizer.py rename to src/backend/app/algorithms/evaluate/clip/simple_tokenizer.py diff --git a/src/backend/app/algorithms/evaluates/libsvm/COPYRIGHT b/src/backend/app/algorithms/evaluate/libsvm/COPYRIGHT similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/COPYRIGHT rename to src/backend/app/algorithms/evaluate/libsvm/COPYRIGHT diff --git a/src/backend/app/algorithms/evaluates/libsvm/FAQ.html b/src/backend/app/algorithms/evaluate/libsvm/FAQ.html similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/FAQ.html rename to src/backend/app/algorithms/evaluate/libsvm/FAQ.html diff --git a/src/backend/app/algorithms/evaluates/libsvm/Makefile b/src/backend/app/algorithms/evaluate/libsvm/Makefile similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/Makefile rename to src/backend/app/algorithms/evaluate/libsvm/Makefile diff --git a/src/backend/app/algorithms/evaluates/libsvm/Makefile.win b/src/backend/app/algorithms/evaluate/libsvm/Makefile.win similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/Makefile.win rename to src/backend/app/algorithms/evaluate/libsvm/Makefile.win diff --git a/src/backend/app/algorithms/evaluates/libsvm/README b/src/backend/app/algorithms/evaluate/libsvm/README similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/README rename to src/backend/app/algorithms/evaluate/libsvm/README diff --git a/src/backend/app/algorithms/evaluates/libsvm/__init__.py b/src/backend/app/algorithms/evaluate/libsvm/__init__.py similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/__init__.py rename to src/backend/app/algorithms/evaluate/libsvm/__init__.py diff --git a/src/backend/app/algorithms/evaluates/libsvm/heart_scale b/src/backend/app/algorithms/evaluate/libsvm/heart_scale similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/heart_scale rename to src/backend/app/algorithms/evaluate/libsvm/heart_scale diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/Makefile b/src/backend/app/algorithms/evaluate/libsvm/java/Makefile similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/Makefile rename to src/backend/app/algorithms/evaluate/libsvm/java/Makefile diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/libsvm.jar b/src/backend/app/algorithms/evaluate/libsvm/java/libsvm.jar similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/libsvm.jar rename to src/backend/app/algorithms/evaluate/libsvm/java/libsvm.jar diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm.java b/src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm.java similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm.java rename to src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm.java diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm.m4 b/src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm.m4 similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm.m4 rename to src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm.m4 diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm_model.java b/src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm_model.java similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm_model.java rename to src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm_model.java diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm_node.java b/src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm_node.java similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm_node.java rename to src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm_node.java diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm_parameter.java b/src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm_parameter.java similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm_parameter.java rename to src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm_parameter.java diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm_print_interface.java b/src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm_print_interface.java similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm_print_interface.java rename to src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm_print_interface.java diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm_problem.java b/src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm_problem.java similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/libsvm/svm_problem.java rename to src/backend/app/algorithms/evaluate/libsvm/java/libsvm/svm_problem.java diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/svm_predict.java b/src/backend/app/algorithms/evaluate/libsvm/java/svm_predict.java similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/svm_predict.java rename to src/backend/app/algorithms/evaluate/libsvm/java/svm_predict.java diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/svm_scale.java b/src/backend/app/algorithms/evaluate/libsvm/java/svm_scale.java similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/svm_scale.java rename to src/backend/app/algorithms/evaluate/libsvm/java/svm_scale.java diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/svm_toy.java b/src/backend/app/algorithms/evaluate/libsvm/java/svm_toy.java similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/svm_toy.java rename to src/backend/app/algorithms/evaluate/libsvm/java/svm_toy.java diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/svm_train.java b/src/backend/app/algorithms/evaluate/libsvm/java/svm_train.java similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/svm_train.java rename to src/backend/app/algorithms/evaluate/libsvm/java/svm_train.java diff --git a/src/backend/app/algorithms/evaluates/libsvm/java/test_applet.html b/src/backend/app/algorithms/evaluate/libsvm/java/test_applet.html similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/java/test_applet.html rename to src/backend/app/algorithms/evaluate/libsvm/java/test_applet.html diff --git a/src/backend/app/algorithms/evaluates/libsvm/libsvm.so.2 b/src/backend/app/algorithms/evaluate/libsvm/libsvm.so.2 similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/libsvm.so.2 rename to src/backend/app/algorithms/evaluate/libsvm/libsvm.so.2 diff --git a/src/backend/app/algorithms/evaluates/libsvm/matlab/Makefile b/src/backend/app/algorithms/evaluate/libsvm/matlab/Makefile similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/matlab/Makefile rename to src/backend/app/algorithms/evaluate/libsvm/matlab/Makefile diff --git a/src/backend/app/algorithms/evaluates/libsvm/matlab/README b/src/backend/app/algorithms/evaluate/libsvm/matlab/README similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/matlab/README rename to src/backend/app/algorithms/evaluate/libsvm/matlab/README diff --git a/src/backend/app/algorithms/evaluates/libsvm/matlab/libsvmread.c b/src/backend/app/algorithms/evaluate/libsvm/matlab/libsvmread.c similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/matlab/libsvmread.c rename to src/backend/app/algorithms/evaluate/libsvm/matlab/libsvmread.c diff --git a/src/backend/app/algorithms/evaluates/libsvm/matlab/libsvmwrite.c b/src/backend/app/algorithms/evaluate/libsvm/matlab/libsvmwrite.c similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/matlab/libsvmwrite.c rename to src/backend/app/algorithms/evaluate/libsvm/matlab/libsvmwrite.c diff --git a/src/backend/app/algorithms/evaluates/libsvm/matlab/make.m b/src/backend/app/algorithms/evaluate/libsvm/matlab/make.m similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/matlab/make.m rename to src/backend/app/algorithms/evaluate/libsvm/matlab/make.m diff --git a/src/backend/app/algorithms/evaluates/libsvm/matlab/svm_model_matlab.c b/src/backend/app/algorithms/evaluate/libsvm/matlab/svm_model_matlab.c similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/matlab/svm_model_matlab.c rename to src/backend/app/algorithms/evaluate/libsvm/matlab/svm_model_matlab.c diff --git a/src/backend/app/algorithms/evaluates/libsvm/matlab/svm_model_matlab.h b/src/backend/app/algorithms/evaluate/libsvm/matlab/svm_model_matlab.h similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/matlab/svm_model_matlab.h rename to src/backend/app/algorithms/evaluate/libsvm/matlab/svm_model_matlab.h diff --git a/src/backend/app/algorithms/evaluates/libsvm/matlab/svmpredict.c b/src/backend/app/algorithms/evaluate/libsvm/matlab/svmpredict.c similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/matlab/svmpredict.c rename to src/backend/app/algorithms/evaluate/libsvm/matlab/svmpredict.c diff --git a/src/backend/app/algorithms/evaluates/libsvm/matlab/svmtrain.c b/src/backend/app/algorithms/evaluate/libsvm/matlab/svmtrain.c similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/matlab/svmtrain.c rename to src/backend/app/algorithms/evaluate/libsvm/matlab/svmtrain.c diff --git a/src/backend/app/algorithms/evaluates/libsvm/python/Makefile b/src/backend/app/algorithms/evaluate/libsvm/python/Makefile similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/python/Makefile rename to src/backend/app/algorithms/evaluate/libsvm/python/Makefile diff --git a/src/backend/app/algorithms/evaluates/libsvm/python/README b/src/backend/app/algorithms/evaluate/libsvm/python/README similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/python/README rename to src/backend/app/algorithms/evaluate/libsvm/python/README diff --git a/src/backend/app/algorithms/evaluates/libsvm/python/__init__.py b/src/backend/app/algorithms/evaluate/libsvm/python/__init__.py similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/python/__init__.py rename to src/backend/app/algorithms/evaluate/libsvm/python/__init__.py diff --git a/src/backend/app/algorithms/evaluates/libsvm/python/allmodel b/src/backend/app/algorithms/evaluate/libsvm/python/allmodel similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/python/allmodel rename to src/backend/app/algorithms/evaluate/libsvm/python/allmodel diff --git a/src/backend/app/algorithms/evaluates/libsvm/python/brisquequality.py b/src/backend/app/algorithms/evaluate/libsvm/python/brisquequality.py similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/python/brisquequality.py rename to src/backend/app/algorithms/evaluate/libsvm/python/brisquequality.py diff --git a/src/backend/app/algorithms/evaluates/libsvm/python/svm.py b/src/backend/app/algorithms/evaluate/libsvm/python/svm.py similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/python/svm.py rename to src/backend/app/algorithms/evaluate/libsvm/python/svm.py diff --git a/src/backend/app/algorithms/evaluates/libsvm/python/svmutil.py b/src/backend/app/algorithms/evaluate/libsvm/python/svmutil.py similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/python/svmutil.py rename to src/backend/app/algorithms/evaluate/libsvm/python/svmutil.py diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-predict b/src/backend/app/algorithms/evaluate/libsvm/svm-predict similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-predict rename to src/backend/app/algorithms/evaluate/libsvm/svm-predict diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-predict.c b/src/backend/app/algorithms/evaluate/libsvm/svm-predict.c similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-predict.c rename to src/backend/app/algorithms/evaluate/libsvm/svm-predict.c diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-scale b/src/backend/app/algorithms/evaluate/libsvm/svm-scale similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-scale rename to src/backend/app/algorithms/evaluate/libsvm/svm-scale diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-scale.c b/src/backend/app/algorithms/evaluate/libsvm/svm-scale.c similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-scale.c rename to src/backend/app/algorithms/evaluate/libsvm/svm-scale.c diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/Makefile b/src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/Makefile similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/Makefile rename to src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/Makefile diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/callbacks.cpp b/src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/callbacks.cpp similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/callbacks.cpp rename to src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/callbacks.cpp diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/callbacks.h b/src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/callbacks.h similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/callbacks.h rename to src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/callbacks.h diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/interface.c b/src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/interface.c similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/interface.c rename to src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/interface.c diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/interface.h b/src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/interface.h similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/interface.h rename to src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/interface.h diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/main.c b/src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/main.c similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/main.c rename to src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/main.c diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/svm-toy.glade b/src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/svm-toy.glade similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-toy/gtk/svm-toy.glade rename to src/backend/app/algorithms/evaluate/libsvm/svm-toy/gtk/svm-toy.glade diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-toy/qt/Makefile b/src/backend/app/algorithms/evaluate/libsvm/svm-toy/qt/Makefile similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-toy/qt/Makefile rename to src/backend/app/algorithms/evaluate/libsvm/svm-toy/qt/Makefile diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-toy/qt/svm-toy.cpp b/src/backend/app/algorithms/evaluate/libsvm/svm-toy/qt/svm-toy.cpp similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-toy/qt/svm-toy.cpp rename to src/backend/app/algorithms/evaluate/libsvm/svm-toy/qt/svm-toy.cpp diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-toy/windows/svm-toy.cpp b/src/backend/app/algorithms/evaluate/libsvm/svm-toy/windows/svm-toy.cpp similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-toy/windows/svm-toy.cpp rename to src/backend/app/algorithms/evaluate/libsvm/svm-toy/windows/svm-toy.cpp diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-train b/src/backend/app/algorithms/evaluate/libsvm/svm-train similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-train rename to src/backend/app/algorithms/evaluate/libsvm/svm-train diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm-train.c b/src/backend/app/algorithms/evaluate/libsvm/svm-train.c similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm-train.c rename to src/backend/app/algorithms/evaluate/libsvm/svm-train.c diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm.cpp b/src/backend/app/algorithms/evaluate/libsvm/svm.cpp similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm.cpp rename to src/backend/app/algorithms/evaluate/libsvm/svm.cpp diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm.def b/src/backend/app/algorithms/evaluate/libsvm/svm.def similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm.def rename to src/backend/app/algorithms/evaluate/libsvm/svm.def diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm.h b/src/backend/app/algorithms/evaluate/libsvm/svm.h similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm.h rename to src/backend/app/algorithms/evaluate/libsvm/svm.h diff --git a/src/backend/app/algorithms/evaluates/libsvm/svm.o b/src/backend/app/algorithms/evaluate/libsvm/svm.o similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/svm.o rename to src/backend/app/algorithms/evaluate/libsvm/svm.o diff --git a/src/backend/app/algorithms/evaluates/libsvm/tools/README b/src/backend/app/algorithms/evaluate/libsvm/tools/README similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/tools/README rename to src/backend/app/algorithms/evaluate/libsvm/tools/README diff --git a/src/backend/app/algorithms/evaluates/libsvm/tools/checkdata.py b/src/backend/app/algorithms/evaluate/libsvm/tools/checkdata.py similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/tools/checkdata.py rename to src/backend/app/algorithms/evaluate/libsvm/tools/checkdata.py diff --git a/src/backend/app/algorithms/evaluates/libsvm/tools/easy.py b/src/backend/app/algorithms/evaluate/libsvm/tools/easy.py similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/tools/easy.py rename to src/backend/app/algorithms/evaluate/libsvm/tools/easy.py diff --git a/src/backend/app/algorithms/evaluates/libsvm/tools/grid.py b/src/backend/app/algorithms/evaluate/libsvm/tools/grid.py similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/tools/grid.py rename to src/backend/app/algorithms/evaluate/libsvm/tools/grid.py diff --git a/src/backend/app/algorithms/evaluates/libsvm/tools/subset.py b/src/backend/app/algorithms/evaluate/libsvm/tools/subset.py similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/tools/subset.py rename to src/backend/app/algorithms/evaluate/libsvm/tools/subset.py diff --git a/src/backend/app/algorithms/evaluates/libsvm/windows/libsvm.dll b/src/backend/app/algorithms/evaluate/libsvm/windows/libsvm.dll similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/windows/libsvm.dll rename to src/backend/app/algorithms/evaluate/libsvm/windows/libsvm.dll diff --git a/src/backend/app/algorithms/evaluates/libsvm/windows/libsvmread.mexw64 b/src/backend/app/algorithms/evaluate/libsvm/windows/libsvmread.mexw64 similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/windows/libsvmread.mexw64 rename to src/backend/app/algorithms/evaluate/libsvm/windows/libsvmread.mexw64 diff --git a/src/backend/app/algorithms/evaluates/libsvm/windows/libsvmwrite.mexw64 b/src/backend/app/algorithms/evaluate/libsvm/windows/libsvmwrite.mexw64 similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/windows/libsvmwrite.mexw64 rename to src/backend/app/algorithms/evaluate/libsvm/windows/libsvmwrite.mexw64 diff --git a/src/backend/app/algorithms/evaluates/libsvm/windows/svm-predict.exe b/src/backend/app/algorithms/evaluate/libsvm/windows/svm-predict.exe similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/windows/svm-predict.exe rename to src/backend/app/algorithms/evaluate/libsvm/windows/svm-predict.exe diff --git a/src/backend/app/algorithms/evaluates/libsvm/windows/svm-scale.exe b/src/backend/app/algorithms/evaluate/libsvm/windows/svm-scale.exe similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/windows/svm-scale.exe rename to src/backend/app/algorithms/evaluate/libsvm/windows/svm-scale.exe diff --git a/src/backend/app/algorithms/evaluates/libsvm/windows/svm-toy.exe b/src/backend/app/algorithms/evaluate/libsvm/windows/svm-toy.exe similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/windows/svm-toy.exe rename to src/backend/app/algorithms/evaluate/libsvm/windows/svm-toy.exe diff --git a/src/backend/app/algorithms/evaluates/libsvm/windows/svm-train.exe b/src/backend/app/algorithms/evaluate/libsvm/windows/svm-train.exe similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/windows/svm-train.exe rename to src/backend/app/algorithms/evaluate/libsvm/windows/svm-train.exe diff --git a/src/backend/app/algorithms/evaluates/libsvm/windows/svmpredict.mexw64 b/src/backend/app/algorithms/evaluate/libsvm/windows/svmpredict.mexw64 similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/windows/svmpredict.mexw64 rename to src/backend/app/algorithms/evaluate/libsvm/windows/svmpredict.mexw64 diff --git a/src/backend/app/algorithms/evaluates/libsvm/windows/svmtrain.mexw64 b/src/backend/app/algorithms/evaluate/libsvm/windows/svmtrain.mexw64 similarity index 100% rename from src/backend/app/algorithms/evaluates/libsvm/windows/svmtrain.mexw64 rename to src/backend/app/algorithms/evaluate/libsvm/windows/svmtrain.mexw64 diff --git a/src/backend/app/algorithms/evaluates/eva_gen_heatmap.py b/src/backend/app/algorithms/evaluates/eva_gen_heatmap.py deleted file mode 100644 index fb27936..0000000 --- a/src/backend/app/algorithms/evaluates/eva_gen_heatmap.py +++ /dev/null @@ -1,520 +0,0 @@ -"""Stable Diffusion 注意力热力图差异可视化工具 (可靠版 - 语义阶段聚合)。 - -本模块使用一种健壮的方法，通过在 Stable Diffusion 扩散模型（U-Net）的 -**早期时间步 (语义阶段)** 捕获并累加交叉注意力权重。这种方法能确保捕获到的 -注意力图信号集中且可靠，用于对比分析干净输入和扰动输入生成的图像对模型 -注意力机制的影响差异。 - -典型用法: - python eva_gen_heatmap.py \\ - --model_path /path/to/sd_model \\ - --image_path_a /path/to/clean_image.png \\ - --image_path_b /path/to/noisy_image.png \\ - --prompt_text "a photo of sks person" \\ - --target_word "sks" \\ - --output_dir output/heatmap_reports -""" - -# 通用参数解析与文件路径管理 -import argparse -import os -from pathlib import Path -from typing import Dict, Any, List, Tuple - -# 数值计算与深度学习依赖 -import torch -import torch.nn.functional as F -import numpy as np -import itertools -import warnings - -# 可视化依赖 -import matplotlib.pyplot as plt -import matplotlib.gridspec as gridspec -from matplotlib.colors import TwoSlopeNorm -from mpl_toolkits.axes_grid1 import make_axes_locatable - -# Diffusers 与 Transformers 依赖 -from diffusers import StableDiffusionPipeline, DPMSolverMultistepScheduler -from diffusers.models.attention_processor import Attention -from transformers import CLIPTokenizer - -# 图像处理与数据读取 -from PIL import Image -from torchvision import transforms - -# 抑制非必要的警告输出 -warnings.filterwarnings("ignore", category=UserWarning) -warnings.filterwarnings("ignore", category=FutureWarning) - - -# ============== 核心模块：注意力捕获与聚合 ============== - -class AttentionMapProcessor: - """自定义注意力处理器，用于捕获 U-Net 交叉注意力层的权重。 - - 通过替换原始的 `Attention` 模块处理器，该类在模型前向传播过程中， - 将所有交叉注意力层的注意力权重（`attention_probs`）捕获并存储。 - - Attributes: - attention_maps (Dict[str, List[torch.Tensor]]): 存储捕获到的注意力图， - 键为层名称，值为该层在不同时间步捕获到的注意力图列表。 - pipeline (StableDiffusionPipeline): 正在处理的 Stable Diffusion 管线。 - original_processors (Dict[str, Any]): 存储原始的注意力处理器，用于恢复。 - current_layer_name (Optional[str]): 当前正在处理的注意力层的名称。 - """ - - def __init__(self, pipeline: StableDiffusionPipeline): - """初始化注意力处理器。 - - Args: - pipeline: Stable Diffusion 模型管线实例。 - """ - self.attention_maps: Dict[str, List[torch.Tensor]] = {} - self.pipeline = pipeline - self.original_processors = {} - self.current_layer_name = None - self._set_processors() - - def __call__( - self, - attn: Attention, - hidden_states: torch.Tensor, - encoder_hidden_states: torch.Tensor = None, - attention_mask: torch.Tensor = None - ) -> torch.Tensor: - """重载 __call__ 方法，执行注意力计算并捕获权重。 - - 此方法替代了原始的 `Attention.processor`，在计算交叉注意力时进行捕获。 - - Args: - attn: 当前的 `Attention` 模块实例。 - hidden_states: U-Net 隐状态 (query)。 - encoder_hidden_states: 文本编码器输出 (key/value)，即交叉注意力输入。 - attention_mask: 注意力掩码。 - - Returns: - 计算后的输出隐状态。 - """ - # 如果不是交叉注意力（即 encoder_hidden_states 为 None），则调用原始处理器 - if encoder_hidden_states is None: - return attn.processor( - attn, hidden_states, encoder_hidden_states, attention_mask - ) - - # 1. 计算 Q, K, V - query = attn.to_q(hidden_states) - key = attn.to_k(encoder_hidden_states) - value = attn.to_v(encoder_hidden_states) - - # 2. 准备矩阵乘法 - query = attn.head_to_batch_dim(query) - key = attn.head_to_batch_dim(key) - - # 3. 计算 Attention Scores (Q @ K^T) - attention_scores = torch.baddbmm( - torch.empty( - query.shape[0], query.shape[1], key.shape[1], - dtype=query.dtype, device=query.device - ), - query, - key.transpose(1, 2), - beta=0, - alpha=attn.scale, - ) - - # 4. 计算 Attention Probabilities - attention_probs = attention_scores.softmax(dim=-1) - layer_name = self.current_layer_name - - # 5. 存储捕获的注意力图 - if layer_name not in self.attention_maps: - self.attention_maps[layer_name] = [] - - # 存储当前时间步的注意力权重 - self.attention_maps[layer_name].append(attention_probs.detach().cpu()) - - # 6. 计算输出 (Attention @ V) - value = attn.head_to_batch_dim(value) - hidden_states = torch.bmm(attention_probs, value) - hidden_states = attn.batch_to_head_dim(hidden_states) - - # 7. 输出层 - hidden_states = attn.to_out[0](hidden_states) - hidden_states = attn.to_out[1](hidden_states) - - return hidden_states - - def _set_processors(self): - """注册自定义处理器，捕获 U-Net 中所有交叉注意力层的权重。 - - 遍历 U-Net 的所有子模块，找到所有交叉注意力层（`Attention` 且名称包含 `attn2`）， - 并将其处理器替换为当前的实例。 - """ - for name, module in self.pipeline.unet.named_modules(): - if isinstance(module, Attention) and 'attn2' in name: - # 存储原始处理器以便后续恢复 - self.original_processors[name] = module.processor - - # 定义一个新的闭包函数，用于在调用前设置当前层的名称 - def set_layer_name(current_name): - def new_call(*args, **kwargs): - self.current_layer_name = current_name - return self.__call__(*args, **kwargs) - return new_call - - module.processor = set_layer_name(name) - - def remove(self): - """恢复 U-Net 的原始注意力处理器，清理钩子。""" - for name, original_processor in self.original_processors.items(): - module = self.pipeline.unet.get_submodule(name) - module.processor = original_processor - self.attention_maps = {} - - -def aggregate_word_attention( - attention_maps: Dict[str, List[torch.Tensor]], - tokenizer: CLIPTokenizer, - target_word: str, - input_ids: torch.Tensor -) -> np.ndarray: - """聚合所有层和语义时间步中目标词汇的注意力图，并进行归一化。 - - 聚合步骤： - 1. 识别目标词汇对应的 Token 索引。 - 2. 对每个层：将所有捕获时间步的注意力图求平均。 - 3. 提取目标 Token 对应的注意力子图，并对 Token 维度求和，对 Attention Heads 求平均。 - 4. 将不同分辨率的注意力图上采样到统一尺寸（64x64）。 - 5. 对所有层的结果进行累加（求和）。 - 6. 最终归一化到 [0, 1]。 - - Args: - attention_maps: 包含各层和时间步捕获的注意力图的字典。 - tokenizer: CLIP 分词器实例。 - target_word: 需要聚焦的关键词。 - input_ids: Prompt 对应的 Token ID 张量。 - - Returns: - 最终聚合并上采样到 64x64 尺寸的注意力热力图 (NumPy 数组)。 - - Raises: - ValueError: 如果无法在 Prompt 中找到目标词汇。 - RuntimeError: 如果未捕获到任何注意力数据。 - """ - - # 1. 识别目标词汇的 Token 索引 - prompt_tokens = tokenizer.convert_ids_to_tokens( - input_ids.squeeze().cpu().tolist() - ) - target_lower = target_word.lower() - target_indices = [] - - for i, token in enumerate(prompt_tokens): - cleaned_token = token.replace('Ġ', '').replace('_', '').lower() - # 查找目标词汇或以目标词汇开头的 token 索引，并排除特殊 token - if (input_ids.squeeze()[i] not in tokenizer.all_special_ids and - (target_lower in cleaned_token or - cleaned_token.startswith(target_lower))): - target_indices.append(i) - - if not target_indices: - print(f"[WARN] 目标词汇 '{target_word}' 未识别。请检查 Prompt 或 Target Word。") - raise ValueError("无法识别目标词汇的 token 索引。") - - # 2. 聚合逻辑 - all_attention_data = [] - # U-Net 输出的最大分辨率（64x64），总像素点数 - TARGET_SPATIAL_SIZE = 4096 - TARGET_MAP_SIZE = 64 - - for layer_name, step_maps in attention_maps.items(): - if not step_maps: - continue - - # 对该层捕获的所有时间步求平均，形状: (batch, heads, spatial_res, target_tokens_len) - avg_map_over_time = torch.stack(step_maps).mean(dim=0) - - # 移除批次维度 (假设 batch size = 1)，形状: (heads, spatial_res, target_tokens_len) - attention_map = avg_map_over_time.squeeze(0) - - # 提取目标 token 的注意力图。形状: (heads, spatial_res, target_indices_len) - target_token_maps = attention_map[:, :, target_indices] - - # 对目标 token 求和 (dim=-1)，对注意力头求平均 (dim=0)，形状: (spatial_res,) - aggregated_map_flat = target_token_maps.sum(dim=-1).mean(dim=0).float() - - # 3. 跨分辨率上采样 - if aggregated_map_flat.shape[0] != TARGET_SPATIAL_SIZE: - # 当前图的尺寸：16x16 (256) 或 32x32 (1024) - map_size = int(np.sqrt(aggregated_map_flat.shape[0])) - map_2d = aggregated_map_flat.reshape(map_size, map_size) - map_to_interp = map_2d.unsqueeze(0).unsqueeze(0) # [1, 1, H, W] - - # 使用双线性插值上采样到 64x64 - resized_map_2d = F.interpolate( - map_to_interp, - size=(TARGET_MAP_SIZE, TARGET_MAP_SIZE), - mode='bilinear', - align_corners=False - ) - resized_map_flat = resized_map_2d.squeeze().flatten() - all_attention_data.append(resized_map_flat) - else: - # 如果已经是 64x64，直接使用 - all_attention_data.append(aggregated_map_flat) - - if not all_attention_data: - raise RuntimeError("未捕获到注意力数据。可能模型或参数设置有误。") - - # 4. 对所有层的结果进行累加 (求和) - final_map_flat = torch.stack(all_attention_data).sum(dim=0).cpu().numpy() - - # 5. 最终归一化到 [0, 1] - final_map_flat = final_map_flat / (final_map_flat.max() + 1e-6) - - map_size = int(np.sqrt(final_map_flat.shape[0])) - final_map_np = final_map_flat.reshape(map_size, map_size) # 64x64 - - return final_map_np - - -def get_attention_map_from_image( - pipeline: StableDiffusionPipeline, - image_path: str, - prompt_text: str, - target_word: str -) -> Tuple[Image.Image, np.ndarray]: - """执行多时间步前向传播，捕获指定图片和 Prompt 的注意力图。 - - 通过只运行扩散过程中的语义阶段（早期时间步）来确保捕获到的注意力权重 - 具有高信号质量。 - - Args: - pipeline: Stable Diffusion 模型管线实例。 - image_path: 待处理的输入图片路径。 - prompt_text: 用于生成图片的 Prompt 文本。 - target_word: 需要聚焦和可视化的关键词。 - - Returns: - 包含 (原始图片, 最终上采样后的注意力图) 的元组。 - """ - print(f"\n-> 正在处理图片: {Path(image_path).name}") - image = Image.open(image_path).convert("RGB").resize((512, 512)) - image_transform = transforms.Compose([ - transforms.ToTensor(), - transforms.Normalize([0.5], [0.5]), - ]) - image_tensor = ( - image_transform(image) - .unsqueeze(0) - .to(pipeline.device) - .to(pipeline.unet.dtype) - ) - - # 1. 编码到 Latent 空间 - with torch.no_grad(): - latent = ( - pipeline.vae.encode(image_tensor).latent_dist.sample() * - pipeline.vae.config.scaling_factor - ) - - # 2. 编码 Prompt - text_input = pipeline.tokenizer( - prompt_text, - padding="max_length", - max_length=pipeline.tokenizer.model_max_length, - truncation=True, - return_tensors="pt" - ) - input_ids = text_input.input_ids - - with torch.no_grad(): - # 获取文本嵌入 - prompt_embeds = pipeline.text_encoder( - input_ids.to(pipeline.device) - )[0] - - # 3. 定义语义时间步 - scheduler = pipeline.scheduler - # 设置扩散步数 (例如 50 步) - scheduler.set_timesteps(50, device=pipeline.device) - - # 只选择语义最丰富的早期 10 步进行捕获 - semantic_steps = scheduler.timesteps[:10] - print(f"-> 正在对语义阶段的 {len(semantic_steps)} 个时间步进行注意力捕获...") - - - processor = AttentionMapProcessor(pipeline) - - try: - # 4. 运行多步 UNet Forward Pass - with torch.no_grad(): - # 在选定的语义时间步上运行 U-Net 预测 - for t in semantic_steps: - pipeline.unet(latent, t, prompt_embeds, return_dict=False) - - # 5. 聚合捕获到的数据 - raw_map_np = aggregate_word_attention( - processor.attention_maps, - pipeline.tokenizer, - target_word, - input_ids - ) - except Exception as e: - print(f"[ERROR] 注意力聚合失败: {e}") - # 确保清理钩子 - raw_map_np = np.zeros(image.size) - finally: - processor.remove() - - # 6. 注意力图上采样到图片尺寸 (512x512) - # PIL 进行上采样 - heat_map_pil = Image.fromarray((raw_map_np * 255).astype(np.uint8)) - heat_map_np_resized = ( - np.array(heat_map_pil.resize( - image.size, - resample=Image.Resampling.LANCZOS # 使用高质量的 Lanczos 滤波器 - )) / 255.0 - ) - - return image, heat_map_np_resized - - -def main(): - """主函数，负责解析参数，加载模型，计算差异并生成可视化报告。""" - parser = argparse.ArgumentParser(description="SD 图片注意力差异可视化报告生成") - parser.add_argument("--model_path", type=str, required=True, - help="Stable Diffusion 模型本地路径。") - parser.add_argument("--image_path_a", type=str, required=True, - help="干净输入图片 (X) 路径。") - parser.add_argument("--image_path_b", type=str, required=True, - help="扰动输入图片 (X') 路径。") - parser.add_argument("--prompt_text", type=str, default="a photo of sks person", - help="用于生成图片的 Prompt 文本。") - parser.add_argument("--target_word", type=str, default="sks", - help="需要在注意力图中聚焦和可视化的关键词。") - parser.add_argument("--output_dir", type=str, default="output", - help="报告 PNG 文件的输出目录。") - args = parser.parse_args() - - print(f"--- 正在生成 Stable Diffusion 注意力差异报告 ---") - - # ---------------- 准备模型 ---------------- - device = 'cuda' if torch.cuda.is_available() else 'cpu' - dtype = torch.float16 if device == 'cuda' else torch.float32 - - try: - # 加载 Stable Diffusion 管线 - pipe = StableDiffusionPipeline.from_pretrained( - args.model_path, - torch_dtype=dtype, - local_files_only=True, - safety_checker=None, - # 从子文件夹加载调度器配置 - scheduler=DPMSolverMultistepScheduler.from_pretrained(args.model_path, subfolder="scheduler") - ).to(device) - except Exception as e: - print(f"[ERROR] 模型加载失败，请检查路径和环境依赖: {e}") - return - - # ---------------- 获取数据 ---------------- - # 获取干净图片 A 的注意力图 M_A - img_A, map_A = get_attention_map_from_image(pipe, args.image_path_a, args.prompt_text, args.target_word) - # 获取扰动图片 B 的注意力图 M_B - img_B, map_B = get_attention_map_from_image(pipe, args.image_path_b, args.prompt_text, args.target_word) - - if map_A.shape != map_B.shape: - print("错误：注意力图尺寸不匹配。中止处理。") - return - - # 计算差异图: Delta = M_A - M_B - diff_map = map_A - map_B - # 计算 L2 范数（差异距离） - l2_diff = np.linalg.norm(diff_map) - print(f"\n计算完毕，注意力图的 L2 范数差异值: {l2_diff:.4f}") - - # ---------------- 绘制专业报告 ---------------- - - # 设置 Matplotlib 字体样式 - plt.rcParams.update({ - 'font.family': 'serif', - 'font.serif': ['DejaVu Serif', 'Times New Roman', 'serif'], - 'mathtext.fontset': 'cm' - }) - - fig = plt.figure(figsize=(12, 16), dpi=120) - - # 3行 x 4列 网格布局，用于图片和图例的精确控制 - gs = gridspec.GridSpec(3, 4, figure=fig, - height_ratios=[1, 1, 1.3], - hspace=0.3, wspace=0.1) - - # --- 第一行：原始图片 --- - ax_img_a = fig.add_subplot(gs[0, 0:2]) - ax_img_b = fig.add_subplot(gs[0, 2:4]) - - # 干净图片 - ax_img_a.imshow(img_A) - ax_img_a.set_title(f"Clean Image ($X$)\nFilename: {Path(args.image_path_a).name}", fontsize=14, pad=10) - ax_img_a.axis('off') - - # 扰动图片 - ax_img_b.imshow(img_B) - ax_img_b.set_title(f"Noisy Image ($X'$)\nFilename: {Path(args.image_path_b).name}", fontsize=14, pad=10) - ax_img_b.axis('off') - - # --- 第二行：注意力热力图 (Jet配色) --- - ax_map_a = fig.add_subplot(gs[1, 0:2]) - ax_map_b = fig.add_subplot(gs[1, 2:4]) - - # 注意力图 A - im_map_a = ax_map_a.imshow(map_A, cmap='jet', vmin=0, vmax=1) - ax_map_a.set_title(f"Attention Heatmap ($M_X$)\nTarget: \"{args.target_word}\"", fontsize=14, pad=10) - ax_map_a.axis('off') - - # 注意力图 B - im_map_b = ax_map_b.imshow(map_B, cmap='jet', vmin=0, vmax=1) - ax_map_b.set_title(f"Attention Heatmap ($M_{{X'}}$)\nTarget: \"{args.target_word}\"", fontsize=14, pad=10) - ax_map_b.axis('off') - - # 为注意力图 B 绘制颜色指示条 - divider = make_axes_locatable(ax_map_b) - cax_map = divider.append_axes("right", size="5%", pad=0.05) - cbar1 = fig.colorbar(im_map_b, cax=cax_map) - cbar1.set_label('Attention Intensity', fontsize=10) - - # --- 第三行：差异对比 (完美居中) --- - # 差异图在网格的中间两列 - ax_diff = fig.add_subplot(gs[2, 1:3]) - - vmax_diff = np.max(np.abs(diff_map)) - # 使用 TwoSlopeNorm 确保 0 值位于色条中央 - norm_diff = TwoSlopeNorm(vmin=-vmax_diff, vcenter=0., vmax=vmax_diff) - - # 使用 Coolwarm 配色，蓝色表示负差异 (M_X' > M_X)，红色表示正差异 (M_X > M_X') - im_diff = ax_diff.imshow(diff_map, cmap='coolwarm', norm=norm_diff) - - title_text = ( - r"Difference Map: $\Delta = M_X - M_{X'}$" + - f"\n$L_2$ Norm Distance: $\mathbf{{{l2_diff:.4f}}}$" - ) - ax_diff.set_title(title_text, fontsize=16, pad=12) - ax_diff.axis('off') - - # 差异图颜色指示条 (居中对齐) - cbar2 = fig.colorbar(im_diff, ax=ax_diff, fraction=0.046, pad=0.04) - cbar2.set_label(r'Scale: Red ($+$) $\leftrightarrow$ Blue ($-$)', fontsize=12) - - # ---------------- 整体修饰与保存 ---------------- - fig.suptitle(f"Museguard: SD Attention Analysis Report", fontsize=20, fontweight='bold', y=0.95) - - output_filename = "heatmap_dif.png" - output_path = Path(args.output_dir) / output_filename - output_path.parent.mkdir(parents=True, exist_ok=True) - - plt.savefig(output_path, bbox_inches='tight', facecolor='white') - print(f"\n专业分析报告已保存至：\n{output_path.resolve()}") - -if __name__ == "__main__": - main() \ No newline at end of file diff --git a/src/backend/app/algorithms/evaluates/eva_gen_nums.py b/src/backend/app/algorithms/evaluates/eva_gen_nums.py deleted file mode 100644 index 31041bd..0000000 --- a/src/backend/app/algorithms/evaluates/eva_gen_nums.py +++ /dev/null @@ -1,513 +0,0 @@ -"""图像生成质量多维度评估工具 (专业重构版)。 - -本脚本用于对比评估两组图像（Clean vs Perturbed）的生成质量。 -支持生成包含指标对比表和深度差异分析的 PNG 报告。 - -Style Guide: Google Python Style Guide -""" - -import os -import time -import subprocess -import tempfile -import warnings -from argparse import ArgumentParser -from pathlib import Path -from typing import Dict, Optional, Tuple, Any - -import torch -import clip -import numpy as np -import pandas as pd -import matplotlib.pyplot as plt -import matplotlib.gridspec as gridspec -from PIL import Image -from torchvision import transforms -from facenet_pytorch import MTCNN, InceptionResnetV1 -from piq import ssim, psnr -import torch_fidelity as fid - -# 抑制非必要的警告输出 -warnings.filterwarnings("ignore", category=UserWarning) -warnings.filterwarnings("ignore", category=FutureWarning) - -# ----------------------------------------------------------------------------- -# 全局配置与样式 -# ----------------------------------------------------------------------------- - -# Matplotlib LaTeX 风格配置 -plt.rcParams.update({ - 'font.family': 'serif', - 'font.serif': ['DejaVu Serif', 'Times New Roman', 'serif'], - 'mathtext.fontset': 'cm', - 'axes.unicode_minus': False -}) - -# 指标元数据配置：定义指标目标方向和分析阈值 -METRIC_ANALYSIS_META = { - 'FID': {'higher_is_better': False, 'th': [2.0, 10.0, 30.0]}, - 'SSIM': {'higher_is_better': True, 'th': [0.01, 0.05, 0.15]}, - 'PSNR': {'higher_is_better': True, 'th': [0.5, 2.0, 5.0]}, - 'FDS': {'higher_is_better': True, 'th': [0.02, 0.05, 0.1]}, - 'CLIP_IQS': {'higher_is_better': True, 'th': [0.01, 0.03, 0.08]}, - 'BRISQUE': {'higher_is_better': False, 'th': [2.0, 5.0, 10.0]}, -} -# 用于综合分析的降级权重 -ANALYSIS_WEIGHTS = {'Severe': 3, 'Significant': 2, 'Slight': 1, 'Negligible': 0} - - -# ----------------------------------------------------------------------------- -# 模型加载 (惰性加载或全局预加载) -# ----------------------------------------------------------------------------- - -try: - CLIP_MODEL, CLIP_PREPROCESS = clip.load('ViT-B/32', 'cuda') - CLIP_MODEL.eval() -except Exception as e: - print(f"[Warning] CLIP 模型加载失败: {e}") - CLIP_MODEL, CLIP_PREPROCESS = None, None - -def _get_clip_text_features(text: str) -> torch.Tensor: - """辅助函数：获取文本的 CLIP 特征。""" - if CLIP_MODEL is None: - return None - tokens = clip.tokenize(text).to('cuda') - with torch.no_grad(): - features = CLIP_MODEL.encode_text(tokens) - features /= features.norm(dim=-1, keepdim=True) - return features - -# ----------------------------------------------------------------------------- -# 核心计算逻辑 -# ----------------------------------------------------------------------------- - -def calculate_metrics( - ref_dir: str, - gen_dir: str, - image_size: int = 512 -) -> Dict[str, float]: - """计算图像集之间的多项质量评估指标。 - - 包括 FDS, SSIM, PSNR, CLIP_IQS, FID。 - - Args: - ref_dir: 参考图片目录路径。 - gen_dir: 生成图片目录路径。 - image_size: 图像处理尺寸。 - - Returns: - 包含各项指标名称和数值的字典。若目录无效返回空字典。 - """ - metrics = {} - - # 1. 数据加载 - def load_images(directory): - imgs = [] - if os.path.exists(directory): - for f in os.listdir(directory): - if f.lower().endswith(('.png', '.jpg', '.jpeg')): - try: - path = os.path.join(directory, f) - imgs.append(Image.open(path).convert("RGB")) - except Exception: - pass - return imgs - - ref_imgs = load_images(ref_dir) - gen_imgs = load_images(gen_dir) - - if not ref_imgs or not gen_imgs: - print(f"[Error] 图片加载失败或目录为空: \nRef: {ref_dir}\nGen: {gen_dir}") - return {} - - device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') - - with torch.no_grad(): - # --- FDS (Face Detection Similarity) --- - print(">>> 计算 FDS...") - mtcnn = MTCNN(image_size=image_size, margin=0, device=device) - resnet = InceptionResnetV1(pretrained='vggface2').eval().to(device) - - def get_face_embeds(img_list): - embeds = [] - for img in img_list: - face = mtcnn(img) - if face is not None: - embeds.append(resnet(face.unsqueeze(0).to(device))) - return torch.stack(embeds) if embeds else None - - ref_embeds = get_face_embeds(ref_imgs) - gen_embeds = get_face_embeds(gen_imgs) - - if ref_embeds is not None and gen_embeds is not None: - # 计算生成集每张脸与参考集所有脸的余弦相似度均值 - sims = [] - for g_emb in gen_embeds: - sim = torch.cosine_similarity(g_emb, ref_embeds).mean() - sims.append(sim) - metrics['FDS'] = torch.tensor(sims).mean().item() - else: - metrics['FDS'] = 0.0 - - # 清理显存 - del mtcnn, resnet - if torch.cuda.is_available(): - torch.cuda.empty_cache() - - # --- SSIM & PSNR --- - print(">>> 计算 SSIM & PSNR...") - tfm = transforms.Compose([ - transforms.Resize((image_size, image_size)), - transforms.ToTensor() - ]) - - # 将参考集堆叠为 [N, C, H, W] - ref_tensor = torch.stack([tfm(img) for img in ref_imgs]).to(device) - - ssim_accum, psnr_accum = 0.0, 0.0 - for img in gen_imgs: - gen_tensor = tfm(img).unsqueeze(0).to(device) # [1, C, H, W] - - # 扩展维度以匹配参考集 - gen_expanded = gen_tensor.expand_as(ref_tensor) - - # 计算单张生成图相对于整个参考集的平均结构相似度 - val_ssim = ssim(gen_expanded, ref_tensor, data_range=1.0) - val_psnr = psnr(gen_expanded, ref_tensor, data_range=1.0) - - ssim_accum += val_ssim.item() - psnr_accum += val_psnr.item() - - metrics['SSIM'] = ssim_accum / len(gen_imgs) - metrics['PSNR'] = psnr_accum / len(gen_imgs) - - # --- CLIP IQS --- - print(">>> 计算 CLIP IQS...") - if CLIP_MODEL: - iqs_accum = 0.0 - txt_feat = _get_clip_text_features("good image") - for img in gen_imgs: - img_tensor = CLIP_PREPROCESS(img).unsqueeze(0).to(device) - img_feat = CLIP_MODEL.encode_image(img_tensor) - img_feat /= img_feat.norm(dim=-1, keepdim=True) - iqs_accum += (img_feat @ txt_feat.T).item() - metrics['CLIP_IQS'] = iqs_accum / len(gen_imgs) - else: - metrics['CLIP_IQS'] = np.nan - - # --- FID --- - print(">>> 计算 FID...") - try: - fid_res = fid.calculate_metrics( - input1=ref_dir, - input2=gen_dir, - cuda=True, - fid=True, - verbose=False - ) - metrics['FID'] = fid_res['frechet_inception_distance'] - except Exception as e: - print(f"[Error] FID 计算异常: {e}") - metrics['FID'] = np.nan - - return metrics - - -def run_brisque_cleanly(img_dir: str) -> float: - """使用 subprocess 和临时目录优雅地执行外部 BRISQUE 脚本。 - - Args: - img_dir: 图像目录路径。 - - Returns: - BRISQUE 分数，若失败返回 NaN。 - """ - print(f">>> 计算 BRISQUE (External)...") - - script_path = Path(__file__).parent / 'libsvm' / 'python' / 'brisquequality.py' - if not script_path.exists(): - print(f"[Error] 找不到 BRISQUE 脚本: {script_path}") - return np.nan - - abs_img_dir = os.path.abspath(img_dir) - - with tempfile.TemporaryDirectory() as temp_dir: - try: - cmd = [ - "python", str(script_path), - abs_img_dir, - temp_dir - ] - - # 在脚本所在目录执行 - subprocess.run( - cmd, - cwd=script_path.parent, - check=True, - stdout=subprocess.PIPE, - stderr=subprocess.PIPE - ) - - # 读取临时生成的日志文件 - log_file = Path(temp_dir) / 'log.txt' - if log_file.exists(): - content = log_file.read_text(encoding='utf-8').strip() - try: - return float(content.split()[-1]) - except ValueError: - return float(content) - else: - return np.nan - - except Exception as e: - print(f"[Error] BRISQUE 执行出错: {e}") - return np.nan - - -# ----------------------------------------------------------------------------- -# 报告可视化与分析逻辑 -# ----------------------------------------------------------------------------- - -def analyze_metric_diff( - metric_name: str, - clean_val: float, - pert_val: float -) -> Tuple[str, str, str]: - """生成科学的分级差异分析文本。 - - Args: - metric_name: 指标名称。 - clean_val: 干净图得分。 - pert_val: 扰动图得分。 - - Returns: - (表头箭头符号, 差异描述文本, 状态等级) - """ - - cfg = METRIC_ANALYSIS_META.get(metric_name) - if not cfg: - return "-", "Configuration not found.", "Negligible" - - diff = pert_val - clean_val - abs_diff = abs(diff) - - # 判定好坏： - is_better = (cfg['higher_is_better'] and diff > 0) or (not cfg['higher_is_better'] and diff < 0) - is_worse = not is_better - - # 确定程度 - th = cfg['th'] - if abs_diff < th[0]: - degree = "Negligible" - elif abs_diff < th[1]: - degree = "Slight" - elif abs_diff < th[2]: - degree = "Significant" - else: - degree = "Severe" - - # 组装文案 - header_arrow = r"$\uparrow$" if cfg['higher_is_better'] else r"$\downarrow$" - - if degree == "Negligible": - analysis_text = f"Negligible change (diff < {th[0]:.4f})." - elif is_worse: - analysis_text = f"{degree} degradation." - else: - analysis_text = f"Unexpected {degree} change." - - return header_arrow, analysis_text, degree - - -def generate_visual_report( - ref_dir: str, - clean_dir: str, - pert_dir: str, - clean_metrics: Dict, - pert_metrics: Dict, - output_path: str -): - """渲染并保存专业对比分析报告 (PNG)。""" - - def get_sample(d): - if not os.path.exists(d): return None, "N/A" - files = [f for f in os.listdir(d) if f.lower().endswith(('.png','.jpg'))] - if not files: return None, "Empty" - return Image.open(os.path.join(d, files[0])).convert("RGB"), files[0] - - img_ref, name_ref = get_sample(ref_dir) - img_clean, name_clean = get_sample(clean_dir) - img_pert, name_pert = get_sample(pert_dir) - - # 布局设置 - # 增加高度以容纳文本 - fig = plt.figure(figsize=(12, 16.5), dpi=120) - gs = gridspec.GridSpec(3, 2, height_ratios=[1, 1, 1.5], hspace=0.25, wspace=0.1) - - # 1. 图像展示区 - ax_ref = fig.add_subplot(gs[0, :]) - if img_ref: - ax_ref.imshow(img_ref) - ax_ref.set_title(f"Reference Image ($X$)\n{name_ref}", fontsize=12, fontweight='bold', pad=10) - ax_ref.axis('off') - - ax_c = fig.add_subplot(gs[1, 0]) - if img_clean: - ax_c.imshow(img_clean) - ax_c.set_title(f"Clean Output ($Y$)\n{name_clean}", fontsize=12, fontweight='bold', pad=10) - ax_c.axis('off') - - ax_p = fig.add_subplot(gs[1, 1]) - if img_pert: - ax_p.imshow(img_pert) - ax_p.set_title(f"Perturbed Output ($Y'$)\n{name_pert}", fontsize=12, fontweight='bold', pad=10) - ax_p.axis('off') - - # 2. 数据表格与分析区 - ax_data = fig.add_subplot(gs[2, :]) - ax_data.axis('off') - - metrics_list = ['FID', 'SSIM', 'PSNR', 'FDS', 'CLIP_IQS', 'BRISQUE'] - table_data = [] - analysis_lines = [] - - degradation_score = 0 - - # 遍历指标生成数据和分析 - for m in metrics_list: - c_val = clean_metrics.get(m, np.nan) - p_val = pert_metrics.get(m, np.nan) - - c_str = f"{c_val:.4f}" if not np.isnan(c_val) else "N/A" - p_str = f"{p_val:.4f}" if not np.isnan(p_val) else "N/A" - diff_str = "-" - - header_arrow = "" - - if not np.isnan(c_val) and not np.isnan(p_val): - # 获取深度分析 - header_arrow, text_desc, degree = analyze_metric_diff(m, c_val, p_val) - - # 计算差异值 - diff = p_val - c_val - # 差异值本身的符号 (Diff > 0 或 Diff < 0) - diff_arrow = r"$\nearrow$" if diff > 0 else r"$\searrow$" - if abs(diff) < 1e-4: diff_arrow = r"$\rightarrow$" - - diff_str = f"{diff:+.4f} {diff_arrow}" - - analysis_lines.append(f"• {m}: Change {diff:+.4f}. Analysis: {text_desc}") - - # 累计降级分数 - cfg = METRIC_ANALYSIS_META.get(m) - is_worse = (cfg['higher_is_better'] and diff < 0) or (not cfg['higher_is_better'] and diff > 0) - if is_worse: - degradation_score += ANALYSIS_WEIGHTS.get(degree, 0) - - # 表格第一列：名称 + 期望方向箭头 - name_with_arrow = f"{m} ({header_arrow})" if header_arrow else m - table_data.append([name_with_arrow, c_str, p_str, diff_str]) - - # 绘制表格 - table = ax_data.table( - cellText=table_data, - colLabels=["Metric (Goal)", "Clean ($Y$)", "Perturbed ($Y'$)", "Diff ($\Delta$)"], - loc='upper center', - cellLoc='center', - colWidths=[0.25, 0.25, 0.25, 0.25] - ) - table.scale(1, 2.0) - table.set_fontsize(11) - - # 美化表头 - for (row, col), cell in table.get_celld().items(): - if row == 0: - cell.set_text_props(weight='bold', color='white') - cell.set_facecolor('#404040') - elif col == 0: - cell.set_text_props(weight='bold') - cell.set_facecolor('#f5f5f5') - - # 3. 底部综合分析文本框 - if not analysis_lines: - analysis_lines.append("• All metrics are missing or invalid.") - - full_text = "Quantitative Difference Analysis:\n" + "\n".join(analysis_lines) - - # 总体结论判断 (基于 holistic degradation score) - conclusion = "\n\n>>> EXECUTIVE SUMMARY (Holistic Judgment):\n" - - if degradation_score >= 8: - conclusion += "CRITICAL DEGRADATION. Significant quality loss observed. Attack highly effective." - elif degradation_score >= 4: - conclusion += "MODERATE DEGRADATION. Observable quality drop in key metrics. Attack effective." - elif degradation_score > 0: - conclusion += "MINOR DEGRADATION. Slight quality loss detected. Attack partially effective." - else: - conclusion += "INEFFECTIVE ATTACK. No significant or unexpected statistical quality loss observed." - - full_text += conclusion - - ax_data.text( - 0.05, - 0.30, - full_text, - ha='left', - va='top', - fontsize=12, family='monospace', wrap=True, - transform=ax_data.transAxes - ) - - fig.suptitle("Museguard: Quality Assurance Report", fontsize=18, fontweight='bold', y=0.95) - - plt.savefig(output_path, bbox_inches='tight', facecolor='white') - print(f"\n[Success] 报告已生成: {output_path}") - - -# ----------------------------------------------------------------------------- -# 主入口 -# ----------------------------------------------------------------------------- - -def main(): - parser = ArgumentParser() - parser.add_argument('--clean_output_dir', type=str, required=True) - parser.add_argument('--perturbed_output_dir', type=str, required=True) - parser.add_argument('--clean_ref_dir', type=str, required=True) - parser.add_argument('--png_output_path', type=str, required=True) - parser.add_argument('--size', type=int, default=512) - args = parser.parse_args() - - - Path(args.png_output_path).parent.mkdir(parents=True, exist_ok=True) - - print("========================================") - print(" Image Quality Evaluation Toolkit") - print("========================================") - - # 1. 计算 Clean 组 - print(f"\n[1/2] Evaluating Clean Set: {os.path.basename(args.clean_output_dir)}") - c_metrics = calculate_metrics(args.clean_ref_dir, args.clean_output_dir, args.size) - if c_metrics: - c_metrics['BRISQUE'] = run_brisque_cleanly(args.clean_output_dir) - - # 2. 计算 Perturbed 组 - print(f"\n[2/2] Evaluating Perturbed Set: {os.path.basename(args.perturbed_output_dir)}") - p_metrics = calculate_metrics(args.clean_ref_dir, args.perturbed_output_dir, args.size) - if p_metrics: - p_metrics['BRISQUE'] = run_brisque_cleanly(args.perturbed_output_dir) - - # 3. 生成报告 - if c_metrics and p_metrics: - generate_visual_report( - args.clean_ref_dir, - args.clean_output_dir, - args.perturbed_output_dir, - c_metrics, - p_metrics, - args.png_output_path - ) - else: - print("\n[Fatal] 评估数据不完整，中止报告生成。") - -if __name__ == '__main__': - main() \ No newline at end of file diff --git a/src/backend/app/controllers/auth_controller.py b/src/backend/app/controllers/auth_controller.py index dac390c..fa15923 100644 --- a/src/backend/app/controllers/auth_controller.py +++ b/src/backend/app/controllers/auth_controller.py @@ -102,7 +102,7 @@ def login(): return jsonify({'error': '账户已被禁用'}), 401 # 创建访问令牌 - 确保用户ID为字符串类型 - access_token = create_access_token(identity=str(user.id)) + access_token = create_access_token(identity=str(user.user_id)) return jsonify({ 'message': '登录成功', diff --git a/src/backend/app/controllers/task_controller.py b/src/backend/app/controllers/task_controller.py index 3eb4160..4457370 100644 --- a/src/backend/app/controllers/task_controller.py +++ b/src/backend/app/controllers/task_controller.py @@ -143,7 +143,7 @@ def create_perturbation_task(current_user_id): return TaskService.json_error('非法的 flow_id 参数') try: - pending_status = TaskService.ensure_status('pending') + waiting_status = TaskService.ensure_status('waiting') perturb_type = TaskService.require_task_type('perturbation') except ValueError as exc: return TaskService.json_error(str(exc), 500) @@ -153,7 +153,7 @@ def create_perturbation_task(current_user_id): flow_id=flow_id, tasks_type_id=perturb_type.task_type_id, user_id=current_user_id, - tasks_status_id=pending_status.task_status_id, + tasks_status_id=waiting_status.task_status_id, description=data.get('description') ) db.session.add(task) @@ -270,9 +270,14 @@ def create_heatmap_task(current_user_id): if not perturbed_image or perturbed_image.task_id != perturbation_task_id: return TaskService.json_error('扰动图片不存在或不属于该任务') + if perturbed_image.image_type: + image_code = perturbed_image.image_type.image_code + if image_code != 'perturbed': + return TaskService.json_error(f'仅支持加噪图生成热力图，当前图片类型为: {perturbed_image.image_type.image_name}', 400) + try: heatmap_type = TaskService.require_task_type('heatmap') - pending_status = TaskService.ensure_status('pending') + waiting_status = TaskService.ensure_status('waiting') except ValueError as exc: return TaskService.json_error(str(exc), 500) @@ -281,7 +286,7 @@ def create_heatmap_task(current_user_id): flow_id=perturbation_task.flow_id, tasks_type_id=heatmap_type.task_type_id, user_id=current_user_id, - tasks_status_id=pending_status.task_status_id, + tasks_status_id=waiting_status.task_status_id, description=data.get('description') ) db.session.add(task) @@ -311,7 +316,7 @@ def start_heatmap_task(task_id, current_user_id): if not task.heatmap: return TaskService.json_error('热力图任务未配置对应图片', 400) - job_id = TaskService.start_heatmap_task(task_id, task.heatmap.images_id) + job_id = TaskService.start_heatmap_task(task_id) if not job_id: return TaskService.json_error('任务启动失败', 500) return jsonify({'message': '任务已启动', 'job_id': job_id}), 200 @@ -372,7 +377,7 @@ def create_finetune_from_perturbation(current_user_id): return TaskService.json_error('微调配置不存在') try: - pending_status = TaskService.ensure_status('pending') + waiting_status = TaskService.ensure_status('waiting') finetune_type = TaskService.require_task_type('finetune') except ValueError as exc: return TaskService.json_error(str(exc), 500) @@ -382,7 +387,7 @@ def create_finetune_from_perturbation(current_user_id): flow_id=perturbation_task.flow_id, tasks_type_id=finetune_type.task_type_id, user_id=current_user_id, - tasks_status_id=pending_status.task_status_id, + tasks_status_id=waiting_status.task_status_id, description=data.get('description') ) db.session.add(task) @@ -443,7 +448,7 @@ def create_finetune_from_upload(current_user_id): return TaskService.json_error('非法的 flow_id 参数') try: - pending_status = TaskService.ensure_status('pending') + waiting_status = TaskService.ensure_status('waiting') finetune_type = TaskService.require_task_type('finetune') except ValueError as exc: return TaskService.json_error(str(exc), 500) @@ -453,7 +458,7 @@ def create_finetune_from_upload(current_user_id): flow_id=flow_id, tasks_type_id=finetune_type.task_type_id, user_id=current_user_id, - tasks_status_id=pending_status.task_status_id, + tasks_status_id=waiting_status.task_status_id, description=data.get('description') ) db.session.add(task) @@ -487,7 +492,24 @@ def start_finetune_task(task_id, current_user_id): job_id = TaskService.start_finetune_task(task_id) if not job_id: return TaskService.json_error('任务启动失败', 500) - return jsonify({'message': '任务已启动', 'job_id': job_id}), 200 + + # 处理返回的 job_id（可能是单个或多个，用逗号分隔） + if ',' in job_id: + # 基于加噪的微调：返回两个 job_id + job_ids = job_id.split(',') + return jsonify({ + 'message': '微调任务已启动（原图和加噪图）', + 'job_id': job_id, + 'job_ids': job_ids, + 'type': 'perturbation-based' + }), 200 + else: + # 上传图片的微调：返回单个 job_id + return jsonify({ + 'message': '微调任务已启动', + 'job_id': job_id, + 'type': 'uploaded' + }), 200 @task_bp.route('/finetune', methods=['GET']) @@ -545,7 +567,7 @@ def create_evaluate_task(current_user_id): try: evaluate_type = TaskService.require_task_type('evaluate') - pending_status = TaskService.ensure_status('pending') + waiting_status = TaskService.ensure_status('waiting') except ValueError as exc: return TaskService.json_error(str(exc), 500) @@ -554,7 +576,7 @@ def create_evaluate_task(current_user_id): flow_id=finetune_task.flow_id, tasks_type_id=evaluate_type.task_type_id, user_id=current_user_id, - tasks_status_id=pending_status.task_status_id, + tasks_status_id=waiting_status.task_status_id, description=data.get('description') ) db.session.add(task) diff --git a/src/backend/app/database/__init__.py b/src/backend/app/database/__init__.py index e1fa86e..1ddd661 100644 --- a/src/backend/app/database/__init__.py +++ b/src/backend/app/database/__init__.py @@ -130,7 +130,7 @@ class TaskStatus(db.Model): """任务状态表""" __tablename__ = 'task_status' task_status_id = db.Column(Integer, primary_key=True, autoincrement=True) - task_status_code = db.Column(String(50), nullable=False, comment='状态代码 (Pending, Processing, Done, Failed)') + task_status_code = db.Column(String(50), nullable=False, comment='状态代码 (Waiting, Processing, Completed, Failed)') task_status_name = db.Column(String(100), nullable=False) description = db.Column(Text) diff --git a/src/backend/app/services/task_service.py b/src/backend/app/services/task_service.py index e2a7bd1..eb0d92d 100644 --- a/src/backend/app/services/task_service.py +++ b/src/backend/app/services/task_service.py @@ -115,6 +115,13 @@ class TaskService: str(user_id), str(flow_id) ) + + @staticmethod + def get_model_data_path(user_id, flow_id): + """模型数据路径: MODEL_DATA_FOLDER/user_id/flow_id""" + return TaskService._build_path( + Config.MODEL_DATA_FOLDER + ) # ==================== 通用辅助功能 ==================== @@ -354,11 +361,16 @@ class TaskService: logger.warning(f"Could not cancel RQ job: {e}") # 更新数据库状态 - failed_status = TaskStatus.query.filter_by(task_status_code='failed').first() - if failed_status: - task.tasks_status_id = failed_status.task_status_id - task.finished_at = datetime.utcnow() - db.session.commit() + try: + failed_status = TaskStatus.query.filter_by(task_status_code='failed').first() + if failed_status: + task.tasks_status_id = failed_status.task_status_id + task.finished_at = datetime.utcnow() + db.session.commit() + except Exception as e: + db.session.rollback() + logger.error(f"Failed to update task status: {e}") + return False return True @@ -406,7 +418,7 @@ class TaskService: logger.error(f"Perturbation config not found") return None - algorithm_code = pert_config.perturbation_algorithm_code + algorithm_code = pert_config.perturbation_code # 加入RQ队列 from app.workers.perturbation_worker import run_perturbation_task @@ -421,7 +433,7 @@ class TaskService: output_dir=output_dir, class_dir=class_dir, algorithm_code=algorithm_code, - epsilon=pert_config.epsilon, + epsilon=perturbation.perturbation_intensity, job_id=job_id, job_timeout='4h' ) @@ -479,17 +491,19 @@ class TaskService: return None # 检测微调类型：查找相同flow_id的Perturbation任务 - perturbation_tasks = Task.query.filter( + perturb_type = TaskService.require_task_type('perturbation') + sibling_perturbation = Task.query.filter( Task.flow_id == task.flow_id, - Task.tasks_type_id == 1, # perturbation类型 + Task.tasks_type_id == perturb_type.task_type_id, Task.tasks_id != task_id - ).all() + ).first() - has_perturbation = len(perturbation_tasks) > 0 + has_perturbation = sibling_perturbation is not None # 路径配置 input_dir = TaskService.get_original_images_path(user_id, task.flow_id) class_dir = TaskService.get_class_data_path(user_id, task.flow_id) + model_data_dir = TaskService.get_model_data_path(user_id, task.flow_id) if has_perturbation: # 类型1：基于加噪结果的微调 @@ -499,52 +513,66 @@ class TaskService: original_input_dir = TaskService.get_original_images_path(user_id, task.flow_id) perturbed_output_dir = TaskService.get_perturbed_generated_path(user_id, task.flow_id, task_id) original_output_dir = TaskService.get_original_generated_path(user_id, task.flow_id, task_id) + # 获取坐标保存路径（3D可视化） - coords_save_path = TaskService._build_path( + original_coords_save_path = TaskService._build_path( Config.COORDS_SAVE_FOLDER, str(user_id), str(task.flow_id), str(task_id), - 'coords.json' + 'original_coords.json' + ) + + # 获取加噪坐标保存路径（3D可视化） + perturbed_coords_save_path = TaskService._build_path( + Config.COORDS_SAVE_FOLDER, + str(user_id), + str(task.flow_id), + str(task_id), + 'perturbed_coords.json' ) # 加入RQ队列 from app.workers.finetune_worker import run_finetune_task queue = TaskService._get_queue() - job_id = f"ft_{task_id}" + job_id_original = f"ft_{task_id}_original" + job_id_perturbed = f"ft_{task_id}_perturbed" job_original = queue.enqueue( run_finetune_task, task_id=task_id, - finetune_method=ft_config.finetune_method, - tranin_images_dir=original_input_dir, - output_model_dir=original_output_dir, + finetune_method=ft_config.finetune_code, + train_images_dir=original_input_dir, + output_model_dir=model_data_dir, class_dir=class_dir, - coords_save_path=coords_save_path, - validation_images_dir=original_output_dir, + coords_save_path=original_coords_save_path, + validation_output_dir=original_output_dir, is_perturbed=False, custom_params=None, - job_id=job_id, + job_id=job_id_original, job_timeout='8h' ) job_perturbed = queue.enqueue( run_finetune_task, task_id=task_id, - finetune_method=ft_config.finetune_method, - tranin_images_dir=perturbed_input_dir, - output_model_dir=perturbed_output_dir, + finetune_method=ft_config.finetune_code, + train_images_dir=perturbed_input_dir, + output_model_dir=model_data_dir, class_dir=class_dir, - coords_save_path=coords_save_path, - validation_images_dir=perturbed_output_dir, + coords_save_path=perturbed_coords_save_path, + validation_output_dir=perturbed_output_dir, is_perturbed=True, custom_params=None, - job_id=job_id, + job_id=job_id_perturbed, job_timeout='8h' ) + logger.info(f"Finetune task {task_id} enqueued with job_ids {job_id_original}, {job_id_perturbed}") + return f"{job_id_original},{job_id_perturbed}" + else: # 类型2：用户上传图片的微调 logger.info(f"Finetune task {task_id}: type=uploaded") @@ -569,20 +597,20 @@ class TaskService: job = queue.enqueue( run_finetune_task, task_id=task_id, - finetune_method=ft_config.finetune_method, - tranin_images_dir=input_dir, - output_model_dir=uploaded_output_dir, + finetune_method=ft_config.finetune_code, + train_images_dir=input_dir, + output_model_dir=model_data_dir, class_dir=class_dir, coords_save_path=coords_save_path, - validation_images_dir=uploaded_output_dir, + validation_output_dir=uploaded_output_dir, is_perturbed=False, custom_params=None, job_id=job_id, job_timeout='8h' ) - - logger.info(f"Finetune task {task_id} enqueued with job_id {job_id}") - return job_id + + logger.info(f"Finetune task {task_id} enqueued with job_id {job_id}") + return job_id except Exception as e: logger.error(f"Error starting finetune task: {e}") @@ -591,13 +619,12 @@ class TaskService: # ==================== Heatmap 任务 ==================== @staticmethod - def start_heatmap_task(task_id, perturbed_image_id): + def start_heatmap_task(task_id): """ 启动热力图任务 Args: task_id: 任务ID - perturbed_image_id: 扰动图片ID Returns: job_id @@ -615,13 +642,19 @@ class TaskService: logger.error(f"Heatmap task {task_id} not found") return None + # 从heatmap对象获取扰动图片ID + perturbed_image_id = heatmap.images_id + if not perturbed_image_id: + logger.error(f"Heatmap task {task_id} has no associated perturbed image") + return None + # 获取扰动图片信息 perturbed_image = Image.query.get(perturbed_image_id) if not perturbed_image: logger.error(f"Perturbed image {perturbed_image_id} not found") return None - user_id = perturbed_image.user_id + user_id = task.user_id # 获取原图（通过father_id关系） if not perturbed_image.father_id: @@ -633,31 +666,20 @@ class TaskService: logger.error(f"Original image not found") return None - # 构建图片路径 - original_image_path = TaskService._build_path( - Config.ORIGINAL_IMAGES_FOLDER, - str(user_id), - str(task.flow_id), - original_image.image_name + # 构建图片路径（使用 stored_filename） + original_image_path = os.path.join( + TaskService.get_original_images_path(user_id, task.flow_id), + original_image.stored_filename ) - perturbed_image_path = TaskService._build_path( - Config.PERTURBED_IMAGES_FOLDER, - str(user_id), - str(task.flow_id), - perturbed_image.image_name + perturbed_image_path = os.path.join( + TaskService.get_perturbed_images_path(user_id, task.flow_id), + perturbed_image.stored_filename ) # 输出目录 output_dir = TaskService.get_heatmap_path(user_id, task.flow_id, task_id) - # 获取模型路径 - sd_version = AlgorithmConfig.STABLE_DIFFUSION_VERSION - model_path = AlgorithmConfig.SD_MODEL_PATHS.get(sd_version) - if not model_path: - logger.error(f"Model path not found for SD version {sd_version}") - return None - # 加入RQ队列 from app.workers.heatmap_worker import run_heatmap_task @@ -670,7 +692,6 @@ class TaskService: original_image_path=original_image_path, perturbed_image_path=perturbed_image_path, output_dir=output_dir, - model_path=model_path, perturbed_image_id=perturbed_image_id, job_id=job_id, job_timeout='2h' diff --git a/src/backend/app/workers/finetune_worker.py b/src/backend/app/workers/finetune_worker.py index 9800085..0129c9c 100644 --- a/src/backend/app/workers/finetune_worker.py +++ b/src/backend/app/workers/finetune_worker.py @@ -18,7 +18,7 @@ logging.basicConfig( logger = logging.getLogger(__name__) -def run_finetune_task(task_id, finetune_config_id, finetune_method, train_images_dir, +def run_finetune_task(task_id, finetune_method, train_images_dir, output_model_dir, class_dir, coords_save_path, validation_output_dir, is_perturbed=False, custom_params=None): """ @@ -26,7 +26,6 @@ def run_finetune_task(task_id, finetune_config_id, finetune_method, train_images Args: task_id: 任务ID - finetune_config_id: 微调配置ID finetune_method: 微调方法 (dreambooth, lora, textual_inversion) train_images_dir: 训练图片目录 output_model_dir: 模型输出目录 @@ -54,13 +53,9 @@ def run_finetune_task(task_id, finetune_config_id, finetune_method, train_images # 获取微调任务详情 finetune = Finetune.query.filter_by( - tasks_id=task_id, - finetune_configs_id=finetune_config_id + tasks_id=task_id ).first() - if not finetune: - raise ValueError(f"Finetune task ({task_id}, {finetune_config_id}) not found") - # 更新任务状态为处理中 processing_status = TaskStatus.query.filter_by(task_status_code='processing').first() if processing_status: @@ -69,7 +64,6 @@ def run_finetune_task(task_id, finetune_config_id, finetune_method, train_images task.started_at = datetime.utcnow() db.session.commit() - logger.info(f"Starting finetune task {task_id} (config: {finetune_config_id})") logger.info(f"Method: {finetune_method}, is_perturbed: {is_perturbed}") # 从数据库获取数据集类型的提示词 @@ -89,7 +83,7 @@ def run_finetune_task(task_id, finetune_config_id, finetune_method, train_images # 清空输出目录（避免旧文件残留） logger.info(f"Clearing output directories...") - for dir_path in [output_model_dir, validation_output_dir, coords_save_path]: + for dir_path in [output_model_dir, validation_output_dir]: if os.path.exists(dir_path): for item in os.listdir(dir_path): item_path = os.path.join(dir_path, item) @@ -97,6 +91,16 @@ def run_finetune_task(task_id, finetune_config_id, finetune_method, train_images os.unlink(item_path) elif os.path.isdir(item_path): shutil.rmtree(item_path) + + # 清理旧的 coords.json 文件 + if os.path.exists(coords_save_path): + if os.path.isfile(coords_save_path): + os.unlink(coords_save_path) + logger.info(f"Removed old coords file: {coords_save_path}") + + # 确保 coords.json 的父目录存在 + coords_dir = os.path.dirname(coords_save_path) + os.makedirs(coords_dir, exist_ok=True) # 运行真实微调算法 result = _run_real_finetune( @@ -208,10 +212,6 @@ def _run_real_finetune(finetune_method, task_id, train_images_dir, output_model_ else: raise ValueError(f"Unsupported finetune method: {finetune_method}") - # 添加is_perturbed标志 - if is_perturbed: - cmd_args.append("--is_perturbed") - # 添加其他默认参数 for key, value in params.items(): if isinstance(value, bool): @@ -268,29 +268,48 @@ def _run_real_finetune(finetune_method, task_id, train_images_dir, output_model_ if process.returncode != 0: raise RuntimeError(f"Finetune failed with code {process.returncode}. Check log: {log_file}") + + # 清理class_dir和output_model_dir（删除所有文件释放空间） + logger.info(f"Cleaning class_dir and output_model_dir to save disk space...") - # 清理class_dir（参考sh脚本） + # 清理class_dir if finetune_method in ['dreambooth', 'lora']: - logger.info(f"Cleaning class directory: {class_dir}") if os.path.exists(class_dir): + logger.info(f"Removing all files in class_dir: {class_dir}") shutil.rmtree(class_dir) os.makedirs(class_dir) - - # 清理output_model_dir中的非图片文件 - logger.info(f"Cleaning non-image files in output directory: {output_model_dir}") + + # 清理output_model_dir（删除所有模型文件，只保留验证图片在validation_output_dir） if os.path.exists(output_model_dir): - import shutil - image_extensions = {'.jpg', '.jpeg', '.png', '.bmp', '.gif', '.webp', '.tiff'} - for item in os.listdir(output_model_dir): - item_path = os.path.join(output_model_dir, item) - if os.path.isfile(item_path): - _, ext = os.path.splitext(item) - if ext.lower() not in image_extensions: - try: - os.remove(item_path) - logger.info(f"Removed non-image file: {item}") - except Exception as e: - logger.warning(f"Failed to remove {item}: {str(e)}") + logger.info(f"Removing all files in output_model_dir: {output_model_dir}") + shutil.rmtree(output_model_dir) + os.makedirs(output_model_dir) + + logger.info(f"Cleanup completed. Only validation images and coords.json are kept.") + + + + # # 清理class_dir（参考sh脚本） + # if finetune_method in ['dreambooth', 'lora']: + # logger.info(f"Cleaning class directory: {class_dir}") + # if os.path.exists(class_dir): + # shutil.rmtree(class_dir) + # os.makedirs(class_dir) + + # # 清理output_model_dir中的非图片文件 + # logger.info(f"Cleaning non-image files in output directory: {output_model_dir}") + # if os.path.exists(output_model_dir): + # image_extensions = {'.jpg', '.jpeg', '.png', '.bmp', '.gif', '.webp', '.tiff'} + # for item in os.listdir(output_model_dir): + # item_path = os.path.join(output_model_dir, item) + # if os.path.isfile(item_path): + # _, ext = os.path.splitext(item) + # if ext.lower() not in image_extensions: + # try: + # os.remove(item_path) + # logger.info(f"Removed non-image file: {item}") + # except Exception as e: + # logger.warning(f"Failed to remove {item}: {str(e)}") return { 'status': 'success', diff --git a/src/backend/app/workers/heatmap_worker.py b/src/backend/app/workers/heatmap_worker.py index 0337e96..b92d28c 100644 --- a/src/backend/app/workers/heatmap_worker.py +++ b/src/backend/app/workers/heatmap_worker.py @@ -18,7 +18,7 @@ logger = logging.getLogger(__name__) def run_heatmap_task(task_id, original_image_path, perturbed_image_path, - output_dir, model_path, perturbed_image_id=None): + output_dir, perturbed_image_id=None): """ 执行热力图生成任务（仅使用真实算法） @@ -27,7 +27,6 @@ def run_heatmap_task(task_id, original_image_path, perturbed_image_path, original_image_path: 原始图片路径 perturbed_image_path: 扰动图片路径 output_dir: 输出目录 - model_path: Stable Diffusion模型路径 perturbed_image_id: 扰动图片ID（用于建立father关系） Returns: @@ -97,7 +96,7 @@ def run_heatmap_task(task_id, original_image_path, perturbed_image_path, # 运行真实热力图算法 result = _run_real_heatmap( task_id, original_image_path, perturbed_image_path, - prompt_text, target_word, output_dir, model_path + prompt_text, target_word, output_dir ) # 保存热力图文件到数据库 @@ -132,7 +131,7 @@ def run_heatmap_task(task_id, original_image_path, perturbed_image_path, def _run_real_heatmap(task_id, original_image_path, perturbed_image_path, - prompt_text, target_word, output_dir, model_path): + prompt_text, target_word, output_dir): """运行真实热力图算法""" from config.algorithm_config import AlgorithmConfig @@ -145,6 +144,14 @@ def _run_real_heatmap(task_id, original_image_path, perturbed_image_path, if not script_path: raise ValueError("Heatmap script not configured") + + default_params = evaluate_config.get('default_params', {}) + model_path = default_params.get('model_path') + + if not model_path: + raise ValueError(f"{evaluate_config} ?{default_params} ?{model_path} Model path not configured in AlgorithmConfig.EVALUATE_SCRIPTS['heatmap']") + + logger.info(f"Using model path from config: {model_path}") # 构建命令行参数 cmd_args = [ diff --git a/src/backend/app/workers/perturbation_worker.py b/src/backend/app/workers/perturbation_worker.py index 69a2a36..148888a 100644 --- a/src/backend/app/workers/perturbation_worker.py +++ b/src/backend/app/workers/perturbation_worker.py @@ -179,21 +179,21 @@ def _run_real_algorithm(script_path, conda_env, algorithm_code, task_id, f"--instance_data_dir_for_adversarial={input_dir}", f"--output_dir={output_dir}", f"--class_data_dir={class_dir}", - f"--pgd_eps={epsilon}", + f"--pgd_eps={int(epsilon)}", ]) elif algorithm_code == 'caat': # CAAT参数结构 cmd_args.extend([ f"--instance_data_dir={input_dir}", f"--output_dir={output_dir}", - f"--eps={epsilon}", + f"--eps={int(epsilon)}", ]) elif algorithm_code == 'pid': # PID参数结构 cmd_args.extend([ f"--instance_data_dir={input_dir}", f"--output_dir={output_dir}", - f"--eps={epsilon}", + f"--eps={int(epsilon)}", ]) else: raise ValueError(f"Unsupported algorithm code: {algorithm_code}") diff --git a/src/backend/config/algorithm_config.py b/src/backend/config/algorithm_config.py index e95933a..22c0af6 100644 --- a/src/backend/config/algorithm_config.py +++ b/src/backend/config/algorithm_config.py @@ -60,15 +60,15 @@ class AlgorithmConfig: 'enable_xformers_memory_efficient_attention': True, 'instance_prompt': 'a photo of sks person', 'class_prompt': 'a photo of person', - 'num_class_images': 200, + 'num_class_images': 5, 'center_crop': True, 'with_prior_preservation': True, 'prior_loss_weight': 1.0, 'resolution': 384, 'train_batch_size': 1, - 'max_train_steps': 2, - 'max_f_train_steps': 3, - 'max_adv_train_steps': 6, + 'max_train_steps': 1, + 'max_f_train_steps': 1, + 'max_adv_train_steps': 1, 'checkpointing_iterations': 1, 'learning_rate': 5e-7, 'pgd_alpha': 0.005, @@ -84,15 +84,15 @@ class AlgorithmConfig: 'enable_xformers_memory_efficient_attention': True, 'instance_prompt': 'a photo of sks person', 'class_prompt': 'a photo of person', - 'num_class_images': 200, + 'num_class_images': 5, 'center_crop': True, 'with_prior_preservation': True, 'prior_loss_weight': 1.0, 'resolution': 384, 'train_batch_size': 1, 'max_train_steps': 2, - 'max_f_train_steps': 3, - 'max_adv_train_steps': 6, + 'max_f_train_steps': 1, + 'max_adv_train_steps': 1, 'checkpointing_iterations': 1, 'learning_rate': 5e-7, 'pgd_alpha': 0.005, @@ -109,7 +109,7 @@ class AlgorithmConfig: 'resolution': 512, 'learning_rate': 1e-5, 'lr_warmup_steps': 0, - 'max_train_steps': 10, + 'max_train_steps': 2, 'hflip': True, 'mixed_precision': 'bf16', 'alpha': 5e-3 @@ -122,7 +122,7 @@ class AlgorithmConfig: 'default_params': { 'pretrained_model_name_or_path': MODELS_DIR['model2'], 'resolution': 512, - 'max_train_steps': 10, + 'max_train_steps': 2, 'center_crop': True, 'attack_type': 'add-log' } @@ -173,16 +173,16 @@ class AlgorithmConfig: 'learning_rate': 2e-6, 'lr_scheduler': 'constant', 'lr_warmup_steps': 0, - 'num_class_images': 200, - 'max_train_steps': 1000, - 'checkpointing_steps': 500, + 'num_class_images': 5, + 'max_train_steps': 4, + 'checkpointing_steps': 2, 'center_crop': True, 'mixed_precision': 'bf16', 'prior_generation_precision': 'bf16', 'sample_batch_size': 5, 'validation_prompt': 'a photo of sks person', - 'num_validation_images': 10, - 'validation_steps': 500, + 'num_validation_images': 2, + 'validation_steps': 1, 'coords_log_interval': 10 } }, @@ -202,14 +202,14 @@ class AlgorithmConfig: 'learning_rate': 1e-4, 'lr_scheduler': 'constant', 'lr_warmup_steps': 0, - 'num_class_images': 200, - 'max_train_steps': 1000, - 'checkpointing_steps': 500, + 'num_class_images': 1, + 'max_train_steps': 4, + 'checkpointing_steps': 2, 'seed': 0, 'mixed_precision': 'fp16', 'rank': 4, 'validation_prompt': 'a photo of sks person', - 'num_validation_images': 10, + 'num_validation_images': 2, 'coords_log_interval': 10 } }, @@ -228,8 +228,8 @@ class AlgorithmConfig: 'learning_rate': 5e-4, 'lr_scheduler': 'constant', 'lr_warmup_steps': 0, - 'max_train_steps': 1000, - 'checkpointing_steps': 500, + 'max_train_steps': 4, + 'checkpointing_steps': 2, 'seed': 0, 'mixed_precision': 'fp16', 'validation_prompt': 'a photo of sks person', @@ -252,7 +252,7 @@ class AlgorithmConfig: 'virtual_script': None, 'conda_env': CONDA_ENVS['pid'], # 使用与微调相同的环境 'default_params': { - 'pretrained_model_name_or_path': MODELS_DIR['model2'], + 'model_path': MODELS_DIR['model2'], } }, 'numbers': { diff --git a/src/backend/config/settings.py b/src/backend/config/settings.py index b97d171..c38028d 100644 --- a/src/backend/config/settings.py +++ b/src/backend/config/settings.py @@ -68,6 +68,7 @@ class Config: MODEL_PERTURBED_FOLDER = os.path.join(MODEL_OUTPUTS_FOLDER, 'perturbed') # 加噪图的模型生成结果 # 微调训练相关配置 CLASS_DATA_FOLDER = os.path.join(STATIC_ROOT, 'class') # 类别数据目录（用于 prior preservation） + MODEL_DATA_FOLDER = os.path.join(STATIC_ROOT, 'model_data') # 模型数据目录（用于微调训练数据存储） # 可视化与分析配置 EVA_RES_FOLDER = os.path.join(STATIC_ROOT, 'eva_res') # 评估结果根目录 COORDS_SAVE_FOLDER = os.path.join(EVA_RES_FOLDER, 'position') # 3D坐标可视化数据（用于训练轨迹）