2D LiDAR Point Transformer

Overview

• dataset/ — dataset download and helper scripts.
• utils/ — utility functions for LiDAR projection, KMeans clustering, and operations that match LiDAR points to image masks.
• training/ — prepare_data.py and train.py to create a JSON dataset mapping and a simple training loop.
• model/ — minimal PyTorch implementation for LidarPointTransformer.
• assets/ — example images/visualizations (the repo includes a placeholder nuscenes_results.svg).

Requirements

Create a Python virtual environment and install the basic dependencies:

python -m venv venv; .\venv\Scripts\Activate.ps1
pip install -r requirements.txt