Zhuofan Zong
Zhuofan Zong
@HITerStudy We find the performance saturates when using more than 6 encoder or decoder layers for larger models (e.g., Swin-L). So we use 6 layers by default.
@zhangchbin, I have fixed it
The eta time is inaccurate in the beginning. You can use DETR aug config if you want to accelerate training as it's faster than LSJ aug. Besides, you better use...
All the provided model weights do not incorporate the auxiliary mask branch.
I can provide the weights but it may take a while to retrain the model.
Are you referring to R50 or SwinL?
I have implemented `with_mask` in the latest commit
@bluceliuljx @Belinda-great, Hi, thanks for your interest in our work! I am sorry that there are some unsolved model export issues. Our repo is implemented using an older version of...
LSJ aug requires more memory than DETR aug. If you adopt a resolution of 1920x1080, it's better to use the config `co_dino_5scale_r50_1x_coco.py`. Besides, you can enable checkpointing by adding `with_cp=True`...
Do you use DINO-4scale?