RPNplus icon indicating copy to clipboard operation
RPNplus copied to clipboard

My training result is bad, why?

Open WaihinWong opened this issue 7 years ago • 4 comments

I train the model in synthetic_dataset. When I run demo.py with params_10000.npy (instead of model.npy), the result is very bad, even can't detect person. How do you train your model @huangshiyu13 ?

WaihinWong avatar Jul 25 '17 02:07 WaihinWong

If you only train detector on synthetic data, the result will be bad.

huangshiyu13 avatar Jul 25 '17 03:07 huangshiyu13

How should I do? I want to train the model as effective as the model.npy

WaihinWong avatar Jul 25 '17 04:07 WaihinWong

You should fellow the training strategy of our paper: https://arxiv.org/abs/1703.06283

huangshiyu13 avatar Jul 25 '17 07:07 huangshiyu13

Thank you very much!!~

WaihinWong avatar Jul 26 '17 06:07 WaihinWong