If I get the bbox and label of an image with other detector, how do I get the scene grpah via pretrained SGG model?

waxnkw / IETrans-SGG.pytorch

This is the code of ECCV 2022 (Oral) paper "Fine-Grained Scene Graph Generation with Data Transfer".

Other

89 stars 6 forks source link

If I get the bbox and label of an image with other detector, how do I get the scene grpah via pretrained SGG model? #8

Closed taotaotao0412 closed 1 year ago

taotaotao0412 commented 1 year ago

Nice job! I want to get more precious detection result because Faster RCNN with low accurate in some complex scene. I want to get scene graph via image, bbox and label. How should I do. THANKS FOR YOUR HELP!

waxnkw commented 1 year ago

You can modify one line at here to add your labels. Note that your provided labels should be aligned with VG's object class ids. Then modify the MODEL.ROI_RELATION_HEAD.USE_GT_OBJECT_LABEL to True in the VG1800-SGCLS or VG50-SGCLS.

taotaotao0412 commented 1 year ago

OK, Thanks a lot