This is the GitHub Repo for the National AI Student Challenge 2022. We have used PeekingDuck as our core tool for Computer Vision.
To try our app immediately, you can click this link, or you can click this link if the first link is not working. Do note that the second link will warn you that your connection is not private. Click on advanced and proceed to the website.
If you have having trouble connecting, do confirm that the link should be a "https://" link, and not a "http://" link.
You can see the instructions at the instructions page of our site.
Running our server locally requires the installation of git, a Version Control Software (to clone the repository), as well as Python3 to run the code, TensorFlow for GPU integration, and finally, a web browser such as Google Chrome or Safari to view the website.
The server requires a GPU and around 3GB of dedicated GPU memory. It can be run on a CPU, but performance will be much worse. Accuracy of the programme is not guaranteed on a CPU.
Git can be installed by visiting this link.
Python (Version 3.9) can be downloaded via this link.
You can view TensorFlow's guide for installation via this link.
Google Chrome can be downloaded via this link.
This can be done by doing the following:
For Windows:
Go to File Explorer and type cmd, and press Enter.
Then paste the following command into the command line (black screen that pops up):
git clone https://github.com/Yuheng3107/aichallenge
After cloning the git repository, there will be a new directory called aichallenge created.
You can move into that directory by typing cd aichallenge into the command line.
In order to run our app, we highly recommend setting up a virtual environment in order to prevent conflicts in package dependencies. If you followed the TensorFlow tutorial, you should now have a conda virtual environment set up.
Creating a virtual environment:
conda create --name tf python=3.9
Activating the virtual environment:
conda activate tf
Deactivating the virtual environment:
conda deactivate
After this, run the following commands in your command line to to install the required dependencies for our app:
conda install cuda-nvcc
pip install -r requirements.txt
Once the installation is successful, you can proceed to running our app using the terminal.
To run the web server locally, run the following command while in the aichallenge directory:
python3 app.py
If the server is successfully running, the prompt
werkzeug INFO: * Running on http://127.0.0.1:5000 (Press CTRL+C to quit) should appear on your command line, and you can paste http://127.0.0.1:5000 in your web browser to access the application web page.
You may view a guide on how to port forward via this link.