search

MVIG-SJTU / AlphaPose

Real-Time and Accurate Full-Body Multi-Person Pose Estimation&Tracking System
http://mvig.org/research/alphapose.html
Other
7.83k stars 1.96k forks source link

--min_box_area does not work with --pose_track #1200

Open Haiyu-HaiyuZhang opened 5 months ago

Haiyu-HaiyuZhang commented 5 months ago

Firstly, thank you guys for the contribution to the project. I tried to use --pose_flow and --pose_track with --min_box_area, I found that --min_box_area does not work when with --pose_track. Here is my cmd:

python scripts/demo_inference.py \
--cfg configs/coco/resnet/256x192_res50_lr1e-3_2x-dcn.yaml \
--checkpoint pretrained_models/fast_dcn_res50_256x192.pth \
--indir [] \
--outdir [] \
--save_img \
--vis_fast \
--showbox \
--min_box_area 6500 \
--pose_flow

These two pictures are visible outputs for these two different commands, left one is with pose_track and the right one iswith pose_flow, you can see the audience on the left picture has not been filtered by min_box_area. Is there any way to fix this? image