Repository for "FaceBoxes: A CPU Real-time Face Detector with High Accuracy", implemented with Caffe, Python interface.
The results of each procedure in FaceBoxes are contained in result folder. The final results are shown in the flowing.
The average time cost is 0.15s/frame. The result is generated by testing a 720P live video.
The accuracy in FDDB which is higher than 0.9. The model contained in as a pretrain model and improve the result
Inatall caff-ssd.
A imdb dataset is necessray to train a model.
QQ群:673071218