search

facebookresearch / DensePose

A real-time approach for mapping all human pixels of 2D RGB images to a 3D surface-based model of the body
http://densepose.org
Other
6.95k stars 1.29k forks source link

Some frames are missing when using infer_simple.py #166

Open francisr opened 5 years ago

francisr commented 5 years ago

I'm using infer_simple.py like in the documentation:

python2 tools/infer_simple.py \
    --cfg configs/DensePose_ResNet101_FPN_s1x-e2e.yaml \
    --output-dir DensePoseData/infer_out/ \
    --image-ext jpg \
    --wts https://s3.amazonaws.com/densepose/DensePose_ResNet101_FPN_s1x-e2e.pkl \
    DensePoseData/video_frames

But many frames end up missing from the output.
I have a video with about 20,000 frames, and after running the inference, only ~6700 have been written.
When I rerun it some of the missing frames are being written, but not all of them.
I'm not sure how to debug this.

stonyhu commented 5 years ago

Dear Contributors, I also had the same question. Some frames were lost. Could you explain the reason?

boycehbz commented 5 years ago

I also have this problem. Maybe the densepose can't detect the the person in these frames.

stonyhu commented 5 years ago

@boycehbz Exactly, I got that.