p3t2ja9zs
/
FaceDection
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

修正少量错误

Browse Source
master
wufayuan 2 years ago
parent 6dd1be543c
commit 4ab20abf82
5 changed files with 63 additions and 69 deletions
Show Stats Download Patch File Download Diff File

60
FaceDetection/cascade.py
Unescape View File

@ -8,7 +8,6 @@ License : MIT License
""" """
from config import POSITIVE_SAMPLE from config import POSITIVE_SAMPLE
from config import NEGATIVE_SAMPLE from config import NEGATIVE_SAMPLE
from config import TRAINING_IMG_HEIGHT from config import TRAINING_IMG_HEIGHT
@ -30,11 +29,11 @@ import numpy
class Cascade: class Cascade:
def __init__(self, face_dir = "", nonface_dir = "", train = True, limit = 30): def __init__(self, face_dir="", nonface_dir="", train=True, limit=30):
#tot_samples = 0 # tot_samples = 0
self.Face = ImageSet(face_dir, sampleNum = POSITIVE_SAMPLE) self.Face = ImageSet(face_dir, sampleNum=POSITIVE_SAMPLE)
self.nonFace = ImageSet(nonface_dir, sampleNum = NEGATIVE_SAMPLE) self.nonFace = ImageSet(nonface_dir, sampleNum=NEGATIVE_SAMPLE)
tot_samples = self.Face.sampleNum + self.nonFace.sampleNum tot_samples = self.Face.sampleNum + self.nonFace.sampleNum
@ -50,14 +49,14 @@ class Cascade:
if DEBUG_MODEL is True: if DEBUG_MODEL is True:
self._mat = numpy.zeros((self.haar.featuresNum, tot_samples)) self._mat = numpy.zeros((self.haar.featuresNum, tot_samples))
for i in xrange(self.Face.sampleNum): for i in range(self.Face.sampleNum):
featureVec = self.haar.calFeatureForImg(self.Face.images[i]) featureVec = self.haar.calFeatureForImg(self.Face.images[i])
for j in xrange(self.haar.featuresNum): for j in range(self.haar.featuresNum):
self._mat[j][i ] = featureVec[j] self._mat[j][i] = featureVec[j]
for i in xrange(self.nonFace.sampleNum): for i in range(self.nonFace.sampleNum):
featureVec = self.haar.calFeatureForImg(self.nonFace.images[i]) featureVec = self.haar.calFeatureForImg(self.nonFace.images[i])
for j in xrange(self.haar.featuresNum): for j in range(self.haar.featuresNum):
self._mat[j][i + self.Face.sampleNum] = featureVec[j] self._mat[j][i + self.Face.sampleNum] = featureVec[j]
numpy.save(FEATURE_FILE_TRAINING, self._mat) numpy.save(FEATURE_FILE_TRAINING, self._mat)
@ -73,14 +72,13 @@ class Cascade:
assert sampleNum == (POSITIVE_SAMPLE + NEGATIVE_SAMPLE) assert sampleNum == (POSITIVE_SAMPLE + NEGATIVE_SAMPLE)
assert featureNum == FEATURE_NUM assert featureNum == FEATURE_NUM
Label_Face = [+1 for i in xrange(POSITIVE_SAMPLE)] Label_Face = [+1 for i in range(POSITIVE_SAMPLE)]
Label_NonFace = [-1 for i in xrange(NEGATIVE_SAMPLE)] Label_NonFace = [-1 for i in range(NEGATIVE_SAMPLE)]
self._label = numpy.array(Label_Face + Label_NonFace) self._label = numpy.array(Label_Face + Label_NonFace)
self.limit = limit self.limit = limit
self.classifierNum = 0 self.classifierNum = 0
self.strong_classifier = [None for i in xrange(limit)] self.strong_classifier = [None for i in range(limit)]
def train(self): def train(self):
@ -95,7 +93,7 @@ class Cascade:
mat = self._mat mat = self._mat
label = self._label label = self._label
for i in xrange(self.limit): for i in range(self.limit):
if cur_fpr < EXPECTED_FPR: if cur_fpr < EXPECTED_FPR:
break break
@ -103,12 +101,12 @@ class Cascade:
cache_filename = ADABOOST_CACHE_FILE + str(i) cache_filename = ADABOOST_CACHE_FILE + str(i)
if os.path.isfile(cache_filename): if os.path.isfile(cache_filename):
self.strong_classifier[i] = getCachedAdaBoost(mat = self._mat, self.strong_classifier[i] = getCachedAdaBoost(mat=self._mat,
label = self._label, label=self._label,
filename= cache_filename, filename=cache_filename,
limit = ADABOOST_LIMIT) limit=ADABOOST_LIMIT)
else: else:
self.strong_classifier[i] = AdaBoost(mat, label, limit = ADABOOST_LIMIT) self.strong_classifier[i] = AdaBoost(mat, label, limit=ADABOOST_LIMIT)
output, fpr = self.strong_classifier[i].train() output, fpr = self.strong_classifier[i].train()
cur_fpr *= fpr cur_fpr *= fpr
@ -120,7 +118,6 @@ class Cascade:
self.classifierNum += 1 self.classifierNum += 1
def updateTrainingDate(self, mat, output, fp_num): def updateTrainingDate(self, mat, output, fp_num):
fp_num = int(fp_num) fp_num = int(fp_num)
@ -131,35 +128,33 @@ class Cascade:
_mat[:, :POSITIVE_SAMPLE] = mat[:, :POSITIVE_SAMPLE] _mat[:, :POSITIVE_SAMPLE] = mat[:, :POSITIVE_SAMPLE]
""" """
for i in xrange(POSITIVE_SAMPLE): for i in range(POSITIVE_SAMPLE):
for j in xrange(FEATURE_NUM): for j in range(FEATURE_NUM):
mat[j][i] = self._mat[j][i] mat[j][i] = self._mat[j][i]
""" """
counter = 0 counter = 0
# only reserve negative samples which are classified wrong # only reserve negative samples which are classified wrong
for i in xrange(POSITIVE_SAMPLE, self._label.size): for i in range(POSITIVE_SAMPLE, self._label.size):
if output[i] != self._label[i]: if output[i] != self._label[i]:
for j in xrange(FEATURE_NUM): for j in range(FEATURE_NUM):
_mat[j][POSITIVE_SAMPLE + counter] = mat[j][i] _mat[j][POSITIVE_SAMPLE + counter] = mat[j][i]
counter += 1 counter += 1
assert counter == fp_num assert counter == fp_num
Label_Face = [+1 for i in xrange(POSITIVE_SAMPLE)] Label_Face = [+1 for i in range(POSITIVE_SAMPLE)]
Label_NonFace = [-1 for i in xrange(fp_num)] Label_NonFace = [-1 for i in range(fp_num)]
_label = numpy.array(Label_Face + Label_NonFace) _label = numpy.array(Label_Face + Label_NonFace)
return _mat, _label return _mat, _label
def predict(self): def predict(self):
output = numpy.zeros(POSITIVE_SAMPLE + NEGATIVE_SAMPLE, dtype= numpy.float16) output = numpy.zeros(POSITIVE_SAMPLE + NEGATIVE_SAMPLE, dtype=numpy.float16)
for i in xrange(self.classifierNum): for i in range(self.classifierNum):
self.strong_classifier[i].prediction(output, th=0)
self.strong_classifier[i].prediction(mat, th = 0)
"""unfinished""" """unfinished"""
@ -168,4 +163,3 @@ class Cascade:
def is_goodenough(self): def is_goodenough(self):
pass pass

12
FaceDetection/image.py
Unescape View File

@ -53,15 +53,15 @@ class Image:
iImg = numpy.zeros((row, col)) iImg = numpy.zeros((row, col))
""" """
for i in xrange(0, row): for i in range(0, row):
for j in xrange(0, col): for j in range(0, col):
if j == 0: if j == 0:
iImg[i][j] = image[i][j] iImg[i][j] = image[i][j]
else: else:
iImg[i][j] = iImg[i][j - 1] + image[i][j] iImg[i][j] = iImg[i][j - 1] + image[i][j]
for j in xrange(0, col): for j in range(0, col):
for i in xrange(1, row): for i in range(1, row):
iImg[i][j] += iImg[i - 1][j] iImg[i][j] += iImg[i - 1][j]
""" """
@ -81,8 +81,8 @@ class Image:
What image.sum() do is the same as the following code What image.sum() do is the same as the following code
but more faster than this. but more faster than this.
for i in xrange(self.Row): for i in range(self.Row):
for j in xrange(self.Col): for j in range(self.Col):
sigma += image[i][j] sigma += image[i][j]
""" """
# sigma = image.sum() # sigma = image.sum()

14
FaceDetection/mapReduce.py
Unescape View File

@ -36,7 +36,7 @@ def map(Face, NonFace):
images_num= len(images) images_num= len(images)
processes = [] processes = []
for i in xrange(PROCESS_NUM): for i in range(PROCESS_NUM):
start = int((i *1./PROCESS_NUM) * images_num) start = int((i *1./PROCESS_NUM) * images_num)
end = int(((i+1)*1./PROCESS_NUM) * images_num ) end = int(((i+1)*1./PROCESS_NUM) * images_num )
sub_imgs = images[start:end] sub_imgs = images[start:end]
@ -46,10 +46,10 @@ def map(Face, NonFace):
FEATURE_FILE_SUBSET + str(i) + ".cache")) FEATURE_FILE_SUBSET + str(i) + ".cache"))
processes.append(process) processes.append(process)
for i in xrange(PROCESS_NUM): for i in range(PROCESS_NUM):
processes[i].start() processes[i].start()
for i in xrange(PROCESS_NUM): for i in range(PROCESS_NUM):
processes[i].join() processes[i].join()
@ -60,7 +60,7 @@ def reduce():
mats = [] mats = []
tot_samples = 0 tot_samples = 0
for i in xrange(PROCESS_NUM): for i in range(PROCESS_NUM):
sub_mat = numpy.load(FEATURE_FILE_SUBSET + str(i) + ".cache" + ".npy") sub_mat = numpy.load(FEATURE_FILE_SUBSET + str(i) + ".cache" + ".npy")
mats.append(sub_mat) mats.append(sub_mat)
tot_samples += sub_mat.shape[1] tot_samples += sub_mat.shape[1]
@ -69,9 +69,9 @@ def reduce():
mat = numpy.zeros((haar.featuresNum, tot_samples), numpy.float32) mat = numpy.zeros((haar.featuresNum, tot_samples), numpy.float32)
sample_readed = 0 sample_readed = 0
for i in xrange(PROCESS_NUM): for i in range(PROCESS_NUM):
for m in xrange(mats[i].shape[0]): # feature number for m in range(mats[i].shape[0]): # feature number
for n in xrange(mats[i].shape[1]): # sample number for n in range(mats[i].shape[1]): # sample number
mat[m][n + sample_readed] = mats[i][m][n] mat[m][n + sample_readed] = mats[i][m][n]

4
FaceDetection/mr_routine.py
Unescape View File

@ -33,9 +33,9 @@ def routine(images, filename):
mat = numpy.zeros((haar.featuresNum, tot_samples), dtype = numpy.float32) mat = numpy.zeros((haar.featuresNum, tot_samples), dtype = numpy.float32)
for i in xrange(tot_samples): for i in range(tot_samples):
featureVec = haar.calFeatureForImg(images[i]) featureVec = haar.calFeatureForImg(images[i])
for j in xrange(haar.featuresNum): for j in range(haar.featuresNum):
mat[j][i] = featureVec[j] mat[j][i] = featureVec[j]
numpy.save(filename, mat) numpy.save(filename, mat)

8
FaceDetection/testing.py
Unescape View File

@ -23,14 +23,14 @@ haar = Feature(TRAINING_IMG_WIDTH, TRAINING_IMG_HEIGHT)
mat = numpy.zeros((haar.featuresNum, tot_samples)) mat = numpy.zeros((haar.featuresNum, tot_samples))
for i in xrange(face.sampleNum): for i in range(face.sampleNum):
featureVec = haar.calFeatureForImg(face.images[i]) featureVec = haar.calFeatureForImg(face.images[i])
for j in xrange(haar.featuresNum): for j in range(haar.featuresNum):
mat[j][i ] = featureVec[j] mat[j][i ] = featureVec[j]
for i in xrange(nonFace.sampleNum): for i in range(nonFace.sampleNum):
featureVec = haar.calFeatureForImg(nonFace.images[i]) featureVec = haar.calFeatureForImg(nonFace.images[i])
for j in xrange(haar.featuresNum): for j in range(haar.featuresNum):
mat[j][i + face.sampleNum] = featureVec[j] mat[j][i + face.sampleNum] = featureVec[j]

Write Preview
Loading…
Cancel
Save