rectangle or label doesn't match object totally .

[xianxiaoyu404]

I have successfully load caffe_weights , but accuracy is too low , i don't know why.

[chuanqi305]

@HAN-Seul I‘m working on it and trying some of this:

1. Replace the ReLU layers by ReLU6
2. Use the same padding as tensorflow
3. Change the post processing layers. Now I'm checking the output data layer by layer and the base MobileNetv2 (without SSD) has been finished.

[chuanqi305]

@HAN-Seul Now all the weights checked, but maybe we need to finetune the model to recover the precision because the changes of ReLU6 and padding.