IdemiaProject

This project is carried out in the context of the Artificial Intelligence Masters of TelecomParis.

_{Made with Python}

Project

Development of an AI evaluating the position of a possibly occulted pedestrian from a picture.

How to setup?

• First, clone the repository:

git clone https://github.com/PierreBio/IdemiaProject.git

• Then go to the root of the project:

cd IdemiaProject

• Create a virtual environment:

py -m venv venv

• Activate your environment:

.\venv\Scripts\activate

• Install requirements:

pip install -r requirements.txt

How to launch?

• Once the project is setup, you can launch DataParser script to parse data:

py -m bin.DataParser

• You can launch RunModel script to train a model after parsing data:

py -m bin.RunModel

• You can launch DataCropper script to create cropped images:

py -m bin.DataCropper

How to use your GPU

Download CUDA toolkit : https://developer.nvidia.com/cuda-downloads You can check your version and if the installation was successfull with:

nvcc --version

Install pytorch with CUDA (remove it if it's already installed without):

pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121

Now you can check if it will use your GPU with this python code:

import torch
print(torch.cuda.is_available())

How to download COCO Dataset?

Go to https://cocodataset.org/#download

Download 2017 Train/Val annotations

How to download OCHuman database

Go to https://github.com/liruilong940607/OCHumanApi?tab=readme-ov-file

Download link is at the bottom of page

Fill in your information and click on Send, you will have access to Google Drive

Download files and save to a OCHuman folder

Use bin/OCParser.py to create the csv files

Ressources