face-anti-spoofing
face-anti-spoofing copied to clipboard
About Loss and Accuracy ?
Hi I'm Thai Hoc
I have a question for training model (I try to ROSE dataset which has 7 classes. But I merge all these classes to 2 new class is 'REAL' and 'FAKE'). But accuracy is very low at 50%. I still use all model of your project But It's impossible
I need to your help to improve my accuracy ? Thanks you for reading my question
Have a good day Thai Hoc
The model of tensorflow is hard to convergence,because no pretrained model. I recommend the mxnet model to you. Please read the first line in ReadMe.