Strange visualization result.
Rapisurazurite opened this issue · 1 comments
Rapisurazurite commented
When I executed the visualization command
python3 demo/demo.py --config-file configs/transfiner/mask_rcnn_R_50_FPN_3x.yaml ----input 'demo/sample_imgs/000000018737.jpg' --opts MODEL.WEIGHTS ./pretrained_model/output_3x_transfiner_r50.pth
The result of segmenting the picture 000000018737.jpg in the demo is strange, why is this?
My pytorch version is 1.10, is it a problem caused by the pytorch version?
000000018737.jpg
000000321214.jpg
000000132408.jpg
lkeab commented
That's due to some incorrect bounding box detection results in small backbone. You can eliminate it by setting higher confidence thresholds or change to lager backbone.