MMDetection Submission

In previous rounds, we used to have a Dockerfile for installing the environment. This time I just see a couple of requirement files. Could someone please tell me how do I put CUDA, PyTorch, mmcv requirements which making a submission?

Hi @gaurav_singhal,

We will be releasing a detailed example for MMDetection submission.
Meanwhile, here are the steps for making the submission on your side.

Addition to requirements.txt:

git+https://github.com/open-mmlab/mmdetection.git@v2.3.0

Example Dockerfile for MMDetection:

Hi @shivam could you please share detailed example for MMDetection submission ?

Yea, until now I cannot submit my mmdet json file generate from test.py, the error not that useful for me to debug :sweat_smile:

Hi @sani, can you share the submission ID?

We can check out and share more details on what may have gone wrong (as well as improve error message).

Hi, this is my sub ID, 173027
Thanks

@shivam not sure if you saw my message, add tag so you get the notification

Hi everyone,

MMdetection resources are now released. Check them out here :point_right:

cc: @sani, @gaurav_singhal

Thanks, @shivam, I am now able to make the active submissions.

2 Likes

Hi, @shivam. We have still an issue with active submission using MMDetection colab.
Could you please check the submission #174325?

Logs from gitlab:

Hi @Mykola_Lavreniuk,

It was happening due to the higher model file size.
We have bumped the maximum file size for the challenge and re-running your submission.

Hey,

It wasn’t a problem for me. What were the size limits before and now?

Hi, it was up to ~2G earlier and it can be technically up to 6G now.

Thank you, very much for quick response.
We have not realize that there are some restrictions on the files sizes that are downloaded.
Thus, we try to upload not just best weights to the model, but other weights as well as a lot of redundant staff that are not necessary for submission…)