hustvl / CrossVIS

[ICCV 2021] Crossover Learning for Fast Online Video Instance Segmentation
https://arxiv.org/abs/2104.05970
Other
85 stars 4 forks source link

Can the results of the training 2019 dataset be uploaded to youtubevos cocoapi for evaluation? #18

Closed 15733171319 closed 2 years ago

15733171319 commented 2 years ago

Hello, I am very interested in your code, would like to ask if the result of your training 2019 dataset can be uploaded to youtubevos cocoapi for evaluation?

vealocia commented 2 years ago

Hi, @15733171319! Thanks for your interest in our work. YouTube-VIS 2019 dataset does not provide annotations for custom evaluation. You must upload your submission to online evaluation server in codalab to get the final scores.

15733171319 commented 2 years ago

你好,@15733171319!感谢您对我们的工作感兴趣。 YouTube-VIS 2019 数据集不提供用于自定义评估的注释。您必须将提交的内容上传到 codalab 中的在线评估服务器才能获得最终分数。

Hello, this is the result after I submitted it, what is the reason? WARNING: Your kernel does not support swap limit capabilities or the cgroup is not mounted. Memory limited without swap. Traceback (most recent call last): File "/tmp/codalab/tmpZIKrQX/run/program/evaluate.py", line 58, in res = gts.loadRes(submit_file) File "build/bdist.linux-x86_64/egg/pycocotools/ytvos.py", line 225, in loadRes AssertionError: Results do not correspond to current coco set

vealocia commented 2 years ago

Your upload file's format maybe invalid.

15733171319 commented 2 years ago

alright, thank you very much