facebookresearch / facebookresearch/detectron2

Densepose on video

Open
#3,280 2 comments 0 reactions 0 assignees View on GitHub
densepose
Dominant language
Python
Stars
34.7k
Forks
7.9k
PR merge metrics
No merged PRs in 30d

Description

I was wondering if it is possible to run `densepose` annotations on a mp4 with `detectron2`?

In the ```projects``` folder, you can run `densepose` with ```applynet.py``` but this only works on images. I tried running this commmand

Instructions To Reproduce the Issue and Full Logs :

I created a folder named DensePose in the configs folder and added in the .yaml and got the model id and file as well from the DensePose model zoo linked here: https://github.com/facebookresearch/detectron2/blob/master/projects/DensePose/README.md
```
cd demo

python demo.py --config-file ../configs/DensePose/R_101_FPN_DL_s1x.yaml \
--video-input video.mp4 \
[--other-options]
--opts MODEL.WEIGHTS detectron2://DensePose/R_101_FPN_DL_s1x/165712116/model_final_844d15.pkl
```

with `densepose` weights and annotations but `Detectron2` gives me this error:

```
Non-existent config key: MODEL.DENSEPOSE_ON

```
I know `DensePose` video exists but it is out of date as it uses `caffe2` separated from `pytorch`.

Is this possible or can you not run on video?

My environment:
- Ubuntu 20.04
- CUDA 11.1
- GTX 980 TI
- OpenCV2
- pytorch 1.8.0+cu111

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.