Maskrcnn-benchmark: How to perform the evaluation for a single category?

[mr2901]

❓ Questions and Help

The result of MaskRCNN is all category.
How to perform the evaluation for a single category?

Thanks

[fmassa]

The model is evaluated on all categories that are contained in your dataset.

If you want to modify the evaluation to only display information about one class, I think you can modify
https://github.com/facebookresearch/maskrcnn-benchmark/blob/f8b0118392aae793771cfc6cd869ad0a601d5c92/maskrcnn_benchmark/data/datasets/evaluation/coco/coco_eval.py#L319-L322

and specify catIds in COCOeval so that you select only a subset of classes to evaluate.

I'd recommend having a look at cocoeval.py to figure out how to pass this parameter, should be related to the Params class

I'm closing this as this is not really related to maskrcnn-benchmark, but let me know if you have further questions

[hande6688]

@fmassa
How to label a single category with labelme and instance segmentation? There are cat, cat, cat, cat0, cat1, cat2 or cat-0, cat-1, cat-2, etc. which annotation method is used?thank you!

I use cat-0, cat-1, cat-2 for annotation, and use labelme2coco to convert annotations.json, where the annotation is [], what's the matter?