The reasoning results obtained from the coco-64.1-map model are questionable.

Sense-X / Co-DETR

[ICCV 2023] DETRs with Collaborative Hybrid Assignments Training

MIT License

950 stars 100 forks source link

The reasoning results obtained from the coco-64.1-map model are questionable. #105

Open Zong-my opened 7 months ago

Zong-my commented 7 months ago

The Coco results obtained using the coco-64.1-map model and demo inference script provided by you are incorrect. Are the models you released based on Coco finetuning? And is the inference script in the demo suitable for all models?

TempleX98 commented 7 months ago

Can you show me the evaluation log?