mtcnn-pytorch icon indicating copy to clipboard operation
mtcnn-pytorch copied to clipboard

Onet trained model not working

Open soham24 opened this issue 6 years ago • 3 comments

I have trained onet by following exact same steps on the same dataset. But when I have replaced it with onet from the demo example. it is not working.

soham24 avatar Jan 29 '19 16:01 soham24

I have meet the same problem. P-Net and R-Net work well, but O-Net doesn't detect any faces or landmarks. Have you solve this problem?

Z-Jeff avatar Jan 20 '20 08:01 Z-Jeff

@Z-Jeff I did not debug it further. You can check blazeface pytorch implementation if looking for face detector

soham24 avatar Jan 20 '20 08:01 soham24

I have solve this problem just now. 'boxes' should be the form of [x1, y1, x2, y2], while the data in "./anno_store/anno_train.txt" is [x1, x2, dx, dy]. I have changed the format in 'gen_Pnet_train_data.py' and 'gen_Rnet_train_data.py', but forgotten to modify 'gen_Onet_train_data.py'. After changing the format in 'gen_Onet_train_data.py', the O-Net works.

Z-Jeff avatar Jan 20 '20 10:01 Z-Jeff