Siyang Liu

icon indicating an email [email protected]

icon company Mindray, Parameter R&D Department icon location Shenzhen, China/ Arlington, VA icon location CV/ Face recognition/ Object detection/ Generative adversarial network/ Deep learning optimization

Results 3 comments of Siyang Liu

insoluble nann problem in training (loss = nan)

I met this problem these day but solved, one hint, you could first look at you dataset, double check whether it has 'dirty' data likes no annotations, xmin >= xmax...

where are the pretrained models?

I will upload the pre-trained models of 4, 10, 20 layers sphereface, thank you.

where are the pretrained models?

I have already submitted all of them, you could see them now.