faceboxes implement by pytorch.《FaceBoxes: A CPU Real-time Face Detector with High Accuracy》 This repository is forked lxg2015,and refer to xiongzihua and add the augmentation from SSD,and change the code to pytorch 0.41.
You should download wider face data. Data annotation is like this:
data/all/image01468.jpg 1 119 185 139 139 1
data/all/image01449.jpg 2 9 39 74 74 1 409 93 77 77 1
format:
path/image_name.jpg num_face x y w h 1 x y w h 1