Skip to content

Latest commit

 

History

History
119 lines (80 loc) · 3.9 KB

SUBMISSION.md

File metadata and controls

119 lines (80 loc) · 3.9 KB

Active Participation - User Guide

Prepare your runtime environment

There are three files that help you setup your environment.

  1. Dockerfile
  2. apt.txt
  3. requirements.txt

Dockerfile

If you plan to use GPU, please make sure that you are using an appropriate CUDA and CUDNN versions. You can specify these at the top of your Docerfile.

apt.txt

If there are certain system level packages that you need, you can specify them in your apt.txt. If you are familiar with ubuntu/debian, this is same as installing these packages using apt-get install command.

requirements.txt

You can specify the list of python packages that need to be installed in your requirements.txt.

Please note that we are using apt.txt and requirements.txt in the Dockerfile to install required packages. We believe that this makes it easier for you to add the required packages without much hassle. If you are comfortable with docker, you are free to edit the Dockerfile as needed.

Initial setup

Before you submit to AIcrowd, you need to setup SSH access to our GitLab instance. This is a one-time requirement to setup your repository.

This process involves

  1. Cloning the repository
  2. Replace git origin to point to your personal repository
  3. Setup SSH key

Clone the repository using:

git clone https://github.com/AIcrowd/food-recognition-challenge-starter-kit
cd food-recognition-challenge-starter-kit

Now, you need to point the repository to your personal repository on AIcrowd GitLab.

git remote set-url origin [email protected]:<your-aicrowd-username>/food-recognition-challenge-starter-kit.git

To be able to push your code to GitLab, you should setup SSH keys first. Please follow the instructions here.

Submit to AIcrowd

To submit to AIcrowd, you need to push a tag starting with submission- to GitLab.

Add the changes to git.

git add --all
git commit -m "<brief summary of changes>"

You need to add large files via git-lfs.

git lfs install

# Add all the files larger than 5 MB to LFS
find . -type f -size +5M -exec git lfs migrate import --include={} &> /dev/null \;

For more information on using LFS, please refer uploading large files to GitLab.

Create and push the tag

# You can replace "-initial-version" with something that describes your submission
git tag -am "submission-initial-version" "submission-initial-version"
git lfs push origin master
git push origin master
git push origin submission-initial-version

Monitor progress and score

After you have done the submission, the progress and live scores will be visible on your GitLab repository -> Issues.

Example scores:

The challenge uses the scores marked with ⭐ for the leaderboards.



🛠 Troubleshooting

Q. My submission failed. How do I know what happened?

If you make a submission in debug mode, we provide the outputs from your code on the GitLab issue page corresponding to your submission. To make a submission in debug mode, you need to add "debug": true in your aicrowd.json. Please note that the debug mode submission will not be considered for leaderboard.

Q. My docker builds fail. Can I reproduce this locally?

You can build the images locally by running the following

pip install -U aicrowd-repo2docker
aicrowd-repo2docker .

Q. What is the code entrypoint?

The evaluator will execute predict.py for generating predictions, so please remember to edit it in your submission!

Q. More questions?

In case you have any doubts or need help, you can reach out to us via Challenge Discussions or Discord.