Skip to content

BEV Representation of an Autonomous car using 6 RGB cameras by making use of Stable Diffusion Transformers

License

Notifications You must be signed in to change notification settings

Raghuram-Veeramallu/DiffTransBEV

Repository files navigation

DiffTransBEV

image image

Generating optimized noise-free BEV representaiton of Autonomous Vehicles (Self-driving cars) using Stable Diffusion Transformers (DiTs), Diffusion Models, SwinV2 Transformers and Lift-Splat-Shoot. NuScenes dataset was used for training and validation, while testing environment was simulated on CARLA simulator.

CARLA Simulator Visualization

Installation

This project works best on:

  • Python 3.7

Steps to recreate the environment

  1. Create conda environment for the project
conda create --name av-bev python=3.7
conda activate av-bev
  1. Install PyTorch
conda install pytorch torchvision -c pytorch

Releases

No releases published

Packages

No packages published