Size Wu (吴思泽)

Hi! You can save the keypoints in .json file in the evaluation process. Add _evaluation=dict(res_folder='path/to/save')_ in the cfg file and run [test](https://github.com/open-mmlab/mmpose/blob/master/docs/en/get_started.md#test-a-dataset), the keypoints will be saved under 'path/to/save'.

Hi! Thanks for your question. The data including camera parameters for [demo](https://github.com/open-mmlab/mmpose/blob/7616b62eabdfed9cbd02fbbfa9411ff37e072be7/demo/body3d_multiview_detect_and_regress_img_demo.py#L243) will be automatically downloaded if you do not specify "img_root". The link you provided was the camera paramters...

You can also download the data for demo in this [link](https://download.openmmlab.com/mmpose/demo/panoptic_body3d_demo.tar).

> Could you please also check if this issue exists in other parts of mmpose? Ok.

> Hello, a follow-up to what I wrote above. I tested the demo, now only on data from CMU: > > Checked the 161029_build1, on +200 frames with demo script...

> So it only multiplies the objectness score with the classification score during inference, because I also got the impression reading the paper that the multiplication would happen during the...

> Nope. Maybe you can try [mmdetection3d](https://github.com/open-mmlab/mmdetection3d/tree/master/configs/centerpoint).

The base unit of the translation parameters in CMU panoptic dataset is centimetre and we transform to millimetre in our code. If yours is metre, the value can be way...

I found ` "transformers_version": "4.38.1",` in the `config.json` of VILA-2.7B. Which version of transformers should we use to run VILA-2.7B? By the way, is there any plan to allow loading...

> > @Taylorminer Hi, thanks for using mmpose. This demo script `body3d_multiview_detect_and_regress_img_demo.py` only support CMU Panoptic data format currently, you can find more description about this demo script in [this...