search

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

About custom image inference code #5

Closed park-youngjae closed 1 year ago

park-youngjae commented 2 years ago

Thank you for your wonderful work!

Is there any plan to upload code about custom image inference code(Predcls, SGCls, SGDet)?

Thanks.

waxnkw commented 2 years ago

Thanks for your interest! I am working on using SGG model for other dataset's relation detection. I think I will write this code before September. If the code is updated, I will inform you through this issue.

Meanwhile, I think you can try to do it with the reference to https://github.com/KaihuaTang/Scene-Graph-Benchmark.pytorch/issues?q=custom. The basic way of using our code is similar to theirs.

park-youngjae commented 2 years ago

Thank you for your quick response.

I'll be looking forward to your code update.

waxnkw commented 1 year ago

Updated. Please refer to DEMO.md.

park-youngjae commented 1 year ago

Thank you for responding to my feedback! Is there a demo for precls task too?

waxnkw commented 1 year ago

Sorry, I do not add the predcls demo because normally it is hard to get object labels exactly matched with VG's classes from other detectors (like COCO pre-trained).

If you have a special requirement and do not want to modify code, you can directly use SGCLS mode and modify the predicted labels to your own after getting the scene graph.

If you want to modify code, you can simply modify one line at here to add your own labels.