aoluming

> Hi, I'm trying this repo as well. I have the same question. I found there is a recognition_demo.py over [here](https://github.com/open-mmlab/mmskeleton/blob/master/mmskeleton/processor/recognition_demo.py). I add the following snippet to [pose_demo.yaml](https://github.com/open-mmlab/mmskeleton/blob/master/configs/pose_estimation/pose_demo.yaml) under ./configs/pose_estimation/...

> when I run this command:mmskl configs/utils/build_dataset_example.yaml --gpus 1 > it will show:Load configuration information from configs/utils/build_dataset_example.yaml > [ ] 0/3, elapsed: 0s, ETA: > There was no progress for...

@Farabi-shafkat Thank you for your kind reply! Have you ever tried to train from scratch on C3D? https://github.com/aoluming/Cost_model ,this is my custom architecture, which I follow the paper 'Collaborative Spatiotemporal...

Hello, you are such a modest man and thank you for doing so much for me. I will try this thread in my code. @Farabi-shafkat

@libb999 同学你用的就是c3d么，有尝试用其他模型么，用这篇repo的c3d加他给的pretrain几个迭代acc就上97，从0训练就1%，我觉得很离谱，你觉得可能是哪里出问题了么。

@libb999 我用其他模型也是百分之1，但是我在训练过程中print了输出，发现了一个问题。就是我在网络中不加dropout的话，输出的类索引基本是一个或两个固定值，不论跑多少epoch都是这样。我不知道是网络的问题还是训练代码的问题，或者是data的问题。但是加了pretrain就很高，说明data可能就没问题

同感，但是检查了loss，print了梯度，感觉是没有问题的，是有梯度回传的，但是就是loss降不下去 @libb999