Icecream-blue-sky

> Hi, the implementation is tricky and making it public needs a lot of work. I can tell you roughly how I achieved it and share some pieces of the...

> > How to merge those per-image results and convert into the format that the evaluation code requires? How to align each per-image results with corresponding gt bboxes? I have...

> > How to merge those per-image results and convert into the format that the evaluation code requires? How to align each per-image results with corresponding gt bboxes? I have...

> > Why the result is arranged by img_id (I mean img_ids = sorted(coco.getImgIds()))? Is it because the annotation info from dataset = build_dataset(cfg.data.test) is arranged by img_id? > >...

> > Do you change the tranforms of training_pipeline to be the same as that of testing_pipeline? I just use the training_pipeline. > > No, I didn't change these configurations....

> @Icecream-blue-sky, sorry, it suddenly came to my mind that you need to change this line: `dict(type='RandomFlip', flip_ratio=0.5)` to `dict(type='RandomFlip', flip_ratio=0.0)` so that the image is not flipped. Thanks!

> @Icecream-blue-sky, sorry, it suddenly came to my mind that you need to change this line: `dict(type='RandomFlip', flip_ratio=0.5)` to `dict(type='RandomFlip', flip_ratio=0.0)` so that the image is not flipped. You are...

> One anchor matches one box in a single class. I haven't seen the use of max-score in other anchor-free methods. Why is it better than dividing the predicted bbox...

I mean it is one box, single-class in YOLOX. But one box mulit-classes in RetinaNet, FCOS. When inference, YOLOX will transform each bbox with multi-classes score to bbox with max...

I also find this problem. It's really confused.