Baek-Donghyeon

I comfronted the same problem, and I managed to solve this abiding by the requirement: pytorch > 1.11.0. Why don't you try this one. `conda install pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=11.3...

> > I comfronted the same problem, and I managed to solve this abiding by the requirement: pytorch > 1.11.0. Why don't you try this one. `conda install pytorch==1.12.1 torchvision==0.13.1...

``` def RGB2D(depth_map): b = depth_map[:,:,2].astype(np.float16) g = depth_map[:,:,1].astype(np.float16) r = depth_map[:,:,0].astype(np.float16) d = np.zeros_like(b, dtype=np.float16) rg = np.logical_and(np.logical_and(r >= g, r >= b), g >= b) d[rg] = g[rg]...

As the error message says, the 'detector' module wasn't not detected. I think you change your current directory to '/content/AlphaPose' For example, instead of using `os.chdir`, do `! cd /content/Alphapose`...

In my opinion, it might be look inappropriate to assess accuracy of locations of predicted keypoints or heatmaps. Instead, pose estimation adopts various metrics for assessment such as PDJ, PCK,...

--format in option only offers 'cmu' or 'open' otherwise it'll give coco formatted json results. If you want to save in MP2 format, I think you'll need to modifiy **write_json**...

Have you tried `bash scripts/validate.sh`

다른 분이 보실 수도 있어서 되도록 영어로 작성하려하니 개의치 않으셔도 됩니다. You can download halpe dataset [here](https://github.com/Fang-Haoshu/Halpe-FullBody#download). Or, I think it's better to just use Halpecocotools API by installing `pip...

In order to train custom dataset, you'll need to make your custom dataset file in **alphapose/datasets**. Assuming that your dataset follows coco format, make **your_custom_dataset.py** copying from **mscoco.py** then modify...