crowdcount-mcnn
crowdcount-mcnn copied to clipboard
ht_1/4 I don't understand why resize image to this size?
den = cv2.resize(ht_1,wd_1) den = den*((htwd)/(ht_1wd_1)) And i couldn't understand what is this meaning?
That's probably because it's using two times of max-pooling layer. That's why it needs to resize the sample to its 1/4 size.