siyiding1216

Results 4 comments of siyiding1216

only get 73.8map on voc2007

@zhreshold Any update on this?

Evaluation results on several models are lower than reported ones.

@zhreshold Are you going to retrain the VGG16 based SSD? We could not reproduce the precision using current training config.

RetinaFace pretrained model

Do you have a timeline to finish the work on RetinaFace?

depthwise convolution to speed up

Are you comparing the latency for your model with mobileNet SSD (without depthwise convolution)? But mobileNet SSD could get 4 timers faster in GPU and 10 times faster in CPU...