Feng Li
It is OK. You can also post your 50.8-51.0 AP model here.
You can refer to DINO visualization script as an example https://github.com/IDEA-Research/DINO/blob/main/inference_and_visualization.ipynb
You can refer to [deformable detr](https://github.com/fundamentalvision/Deformable-DETR) for more details about cuda installation, as we use its deformable cuda implementation.
Maybe you did not set this parameter right.
Code link in https://github.com/IDEA-Research/MaskDINO/blob/main/configs/coco/instance-segmentation/Base-COCO-InstanceSegmentation.yaml#:~:text=%23%20not%20used-,DATASETS,-%3A
We did not try onnx before, so currently, we could not offer any suggestions.
For detection, using learnable content query could be better. Mask DINO mainly focuses on segmentation that is deeply related to content query, so we use selected content query.
The code will be released very soon. We are preparing these days to implement it with [detrex](https://github.com/IDEA-Research/detrex) (:
Happy December! Our code is available now!
Hey, it starts from zero.