net = caffe.Net(prototxt, caffemodel, caffe.TEST)
print '\n\nLoaded network {:s}'.format(caffemodel)
# Warmup on a dummy image
im = 128 * np.ones((1000, 512, 3), dtype=np.uint8)
for i in xrange(2):
_, __= im_detect(net, im)
print '\nDetect begin: \n'
timer=Timer()
timer.tic()
for wav_name in sorted(os.listdir(test_folder)):
if wav_name.endswith('.wav'):
full_wav_name=os.path.join(test_folder,wav_name)
x,rate=load_audio(full_wav_name,common_fs)
image=stfft(x,nfft,noverlap)
IMG=Image.fromarray(image.astype(np.uint8))
IMG.save('temp.jpg')
im=cv2.imread('temp.jpg')
os.remove('temp.jpg')
'''
im=np.zeros((image.shape[0],image.shape[1],3))
for k in range(3):
im[:,:,k]=np.uint8(image)
'''
#timer.tic()
scores, boxes = im_detect(net, im)
#timer.toc()
#print ('Detection took {:.3f}s for '
#'{:d} object proposals').format(timer.total_time, boxes.shape[0])
image_num = 0
for item in data:
image_num += 1
full_audio_name = audio + item['mixture_audio_filename']
if os.path.exists(full_audio_name):
x, rate = load_audio(full_audio_name, common_fs)
else:
continue
picture_interest_start = int(
rate * item['event_start_in_mixture_seconds'] /
(nfft - noverlap))
picture_interest_end = int(
rate * (item['event_start_in_mixture_seconds'] +
item['event_length_seconds']) / (nfft - noverlap))
im = stfft(x, nfft, noverlap)
image = Image.fromarray(im.astype(np.uint8))
image_name = 'IMG_' + str(i) + str(
EBR[item['ebr']]) + "%04d" % (image_num)
image.save(data_folder + 'JPEGImages/' + image_name + '.jpg')
fpicture.write(image_name + '.jpg ' + full_audio_name + '\n')
ftrainval.write(image_name + '\n')
if (image_num < int(num_of_every_class * 0.8)):
ftrain.write(image_name + '\n')
else:
fval.write(image_name + '\n')
label_txt = data_folder + 'Labels/' + image_name + '.txt'
flabel = open(label_txt, 'wt')
flabel.write(image_name+'\n' \
+str(image.size[0])+' '+str(image.size[1])+' 1\n')
flabel.write(CLASS_NAMES[i]+'\n' \