AI Singapore Trusted Media Challenge Submission Guide

Participants are to submit a compressed Docker image in the tar.gz format onto the challenge platform. This repository serves as a step by step guide to help participants with creating a valid submission for the Trusted Media Challenge.

Getting Started

We would be using Docker for this challenge so that participants can choose their preferred programming languages and dependencies to create the best perfoming detection models.

To build and run GPU accelerated Docker containers, please install the NVIDIA Container Toolkit.

Usage of sample submission

Clone this repository and navigate to it

git clone https://github.com/AISG-Technology-Team/Trusted-Media-Challenge-Submission-Example.git

Change into the sample submission directory

cd sample_submission

Build the sample Docker image

docker build -t sample_image .

Please take note of the “.” indicates the current project directory and should be added into the docker build command to provide the build context.

Test sample image locally

Add a few mp4 videos into local_test/test_input and replace "/path/to/cloned/repo" in the following command with the full path to the cloned repo. The test is successful if no error messages were seen and a submission.csv is created in the local_test/test_output directory.

docker run --rm -it \
 -v /path/to/cloned/repo/local_test/test_input:/data/input \
 -v /path/to/cloned/repo/local_test/test_output:/data/output \
 sample_image \
 -input /data/input/ -output /data/output/submission.csv

Compress your sample image to .tar.gz format using docker save

docker save sample_image:latest | gzip > sample_image.tar.gz

Upload sample image

Submit your sample_image.tar.gz file onto the challenge platform

Creating your own submission

The process of creating your own submission would be very similar to using the aforementioned sample submission.

Create a project directory and navigate into it

mkdir Trusted-Media-Challenge && cd Trusted-Media-Challenge

Create a main file

The main file has to accept two arguments:

-input is a directory which contains all videos of the test set, e.g. /data/input/ ('/' should appear at the end of the line)
-output is the name (with path) of the output file, e.g. /data/output/submission.csv

and has three main functions:

Read the mp4 videos from directory specified in the aforementioned -input argument
Predicts the probability that each input video is fake
Writes a csv output file with the name and at the path specified in the aforementioned -output argument. The csv file should contain two columns filename and probability where probability is the estimated probability that the video, stated in filename, is fake. An example csv is provided.

You may refer to the main.py of the sample submission as an example of a main file.

Create a Dockerfile

You may use the sample Dockerfile provided for you. However, please install the relevant dependencies required for your detection model. Additionally, you may wish to change the ENTRYPOINT if you are using another main file or if you prefer to use a shell script:

ENTRYPOINT ["bash","/path/to/your/main.sh"]

If you are not familiar with building a Dockerfile, please refer to the official documentation for more information.

Build your Docker image using docker build

docker build -t your_image .

Please take note of the “.” indicates the current project directory and should be added into the docker build command to provide the build context.

Test your image locally

1. Create a test directory outside of your project directory

mkdir local_test && cd local_test

2. Create input and output directory within your test directory

mkdir test_input test_output

3. Add test mp4 videos into test_input directory

4. Test your image using docker run

Please replace "/path/to/test/dir" in the following command with the full path to your test directory. The test is successful if no error messages were seen and a submission.csv is created in the local_test/test_output directory.

docker run --rm -it --gpus ‘”device=0”’ \
 -v /path/to/test/dir/local_test/test_input:/data/input \
 -v /path/to/test/dir/local_test/test_output:/data/output \
 your_image \
 -input /data/input/ -output /data/output/submission.csv

Please note that above docker run command on the sample image would be equivalent to running the following command in the container:

python /app/main.py -input /data/input/ -output /data/output/submission.csv

Compress your Docker image to .tar.gz format using docker save

docker save your_image:latest | gzip > your_image.tar.gz

Upload your image

Submit your your_image.tar.gz file onto the challenge platform

Vanova/Trusted-Media-Challenge-Submission-Guide