luoyetx / Joint-Face-Detection-and-Alignment

Caffe and Python implementation of Joint Face Detection and Alignment using Multi-task Cascaded Convolutional Networks
BSD 3-Clause "New" or "Revised" License
102 stars 50 forks source link

data layer data queue #4

Closed ForestWang closed 6 years ago

ForestWang commented 6 years ago

Hi luoyetx: thank you for your wonderful work and open source! def set_batch_num(self, n1, n2, n3, n4): '''set data type number

n1 for negatives, n2 for positives, n3 for part faces, n4 for landmark faces

net_input_size for network input size (width, height)

def set_data_queue(self, queue): '''the queue should put a minibatch with size of (positives, negatives, part faces, landmark faces) = (n1, n2, n3, n4) in a dict ''' so which one is correct, look through the code, negative is n1,and positive is n2, is it right? Forest

luoyetx commented 6 years ago

some mistakes on code comment, n1 is for negative, view here and here