AI-PROJECT-SMART_RECORDING_PIPELINE

How to use the code ?

If you're facing path issue while running like "the module 'local_test' does not exist" type this command in your terminal to add the project to your python paht :

export PYTHONPATH="${PYTHONPATH}:<PATH_TO_YOUR_PROJECT>"

Your project's data has to have the same tree as the a2d2 dataset tree (example): Inside each cam_front_semantic_bboxes files you'll find specific files from the a2d2 dataset (.npz for lidar, .json/.png from camera)

📦camera_lidar_semantic_bboxes
 ┃ ┣ 📂20181107_132300
 ┃ ┃ ┣ 📂camera
 ┃ ┃ ┃ ┗ 📂cam_front_center
 ┃ ┃ ┣ 📂label
 ┃ ┃ ┃ ┗ 📂cam_front_center
 ┃ ┃ ┣ 📂label3D
 ┃ ┃ ┃ ┗ 📂cam_front_center
 ┃ ┃ ┣ 📂lidar
 ┃ ┃ ┃ ┗ 📂cam_front_center
 ┣ 📜cams_lidar.json
 ┗ 📜class_list.json

To run the code :

WARNING : Do not forget to change the paths in the 3d_detection file

version of python : 3.10.11

CONDA

To create an environnement with conda please use the command :

conda create --name <env> --file requirement_CONDA.txt python=3.10.11

If you're facing issues with the requirement.txt file, you can try to use the yml file running this command :

conda env create -f environment_conda.yml

VENV

After creating your venv envrionement, you can install the packages with pip please use the command :

pip install -r requirements_pip.txt

Then to run the code, go to the root folder of your project and run the command :

python 3d_projection/3d_projection.py

Models :

YOLOPv2 weights are downloadable here.

The pipeline :