def hog_train():
# features = []
# labels = []
svm = svm_config()
# get hog features
features, labels = generate_hog_data(batch_size=10000)
print(features.shape, labels.shape)
# svm training
print('svm training...')
svm_train(svm, features, labels)
print('svm training complete...')
svm_save(svm, "../../models/HOGSVM_Digits.bin")
#hog训练
def hog_train(svm):
features = []
labels = []
hog = cv.HOGDescriptor()
#get hog features
get_features(features, labels)
#svm training
print('svm training...')
svm_train(svm, features, labels)
print('svm training complete...')
hog.setSVMDetector(get_svm_detector(svm))
hog.save('myHogDector.bin')
print('hard samples training...')
get_hard_samples(svm, features, labels)
print('hard samples complete...')
if __name__ == '__main__':
#svm config
svm = svm_config()
#hog training
hog_train(svm)
if __name__ == '__main__':
features = []
labels = []
pos_imgs, pos_labels = read_pos_samples('E:/pypj/res')
get_features(features, labels)
scores = []
result = []
print('start validate')
pbar = tqdm(total=100)
step = (-5, -4, -3, -2, -1, 0, 1, 2, 3, 4, 5)
for i in step:
C = 2**i
for j in step:
G = 2**j
svm = svm_config(C, G)
svm_train(svm, features, labels)
hog = hog_config()
hog.setSVMDetector(get_svm_detector(svm))
for pos_img in pos_imgs:
rects, score = hog.detect(pos_img)
if len(score) == 0:
score = np.array([0])
scores.append(score.mean())
result.append([C, G, np.array(scores).mean()])
scores = []
pbar.update(100 / len(step)**2)
result1 = sorted(result, key=lambda x: x[2])
print('complete!')
for i in result1:
print(i)
