{"id":13445215,"url":"https://github.com/autonomousvision/neat","last_synced_at":"2025-09-09T06:08:42.265Z","repository":{"id":38352640,"uuid":"397601459","full_name":"autonomousvision/neat","owner":"autonomousvision","description":"[ICCV'21] NEAT: Neural Attention Fields for End-to-End Autonomous Driving","archived":false,"fork":false,"pushed_at":"2022-10-22T15:35:54.000Z","size":34131,"stargazers_count":321,"open_issues_count":4,"forks_count":49,"subscribers_count":16,"default_branch":"main","last_synced_at":"2025-05-20T01:11:24.476Z","etag":null,"topics":["autonomous-driving","birds-eye-view","iccv2021","imitation-learning","implicit-representations"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/autonomousvision.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null}},"created_at":"2021-08-18T12:59:21.000Z","updated_at":"2025-05-08T07:30:38.000Z","dependencies_parsed_at":"2023-01-19T15:30:29.583Z","dependency_job_id":null,"html_url":"https://github.com/autonomousvision/neat","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/autonomousvision/neat","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/autonomousvision%2Fneat","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/autonomousvision%2Fneat/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/autonomousvision%2Fneat/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/autonomousvision%2Fneat/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/autonomousvision","download_url":"https://codeload.github.com/autonomousvision/neat/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/autonomousvision%2Fneat/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":274250722,"owners_count":25249440,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-09-09T02:00:10.223Z","response_time":80,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["autonomous-driving","birds-eye-view","iccv2021","imitation-learning","implicit-representations"],"created_at":"2024-07-31T05:00:26.473Z","updated_at":"2025-09-09T06:08:42.239Z","avatar_url":"https://github.com/autonomousvision.png","language":"Python","funding_links":[],"categories":["二、Camera-based BEV","Python"],"sub_categories":["1. List of camera-based BEV sensing methods"],"readme":"# NEAT: Neural Attention Fields for End-to-End Autonomous Driving\n\n## [Paper](http://www.cvlibs.net/publications/Chitta2021ICCV.pdf) | [Supplementary](http://www.cvlibs.net/publications/Chitta2021ICCV_supplementary.pdf) | [Video](https://www.youtube.com/watch?v=gtO-ghjKkRs) | [Talk](https://www.youtube.com/watch?v=hYm6LPTyHHA) | [Poster](https://www.cvlibs.net/publications/Chitta2021ICCV_poster.pdf) | [Slides](https://www.cvlibs.net/publications/Chitta2021ICCV_slides.pdf)\n\n\u003cimg src=\"neat/assets/neat_clip.GIF\" height=\"270\" hspace=30\u003e\n\nThis repository is for the ICCV 2021 paper [NEAT: Neural Attention Fields for End-to-End Autonomous Driving](http://www.cvlibs.net/publications/Chitta2021ICCV.pdf).\n\n```bibtex\n@inproceedings{Chitta2021ICCV,\n  author = {Chitta, Kashyap and Prakash, Aditya and Geiger, Andreas},\n  title = {NEAT: Neural Attention Fields for End-to-End Autonomous Driving},\n  booktitle = {International Conference on Computer Vision (ICCV)},\n  year = {2021}\n}\n```\n\n## Setup\nPlease follow the installation instructions from our [TransFuser repository](https://github.com/autonomousvision/transfuser/tree/cvpr2021) to set up the CARLA simulator. The conda environment required for NEAT can be installed via:\n```Shell\nconda env create -f environment.yml\nconda install pytorch torchvision torchaudio cudatoolkit=11.1 -c pytorch -c nvidia\n```\n\nFor running the AIM-VA baseline, you will additionally need to install [MMCV](https://mmcv.readthedocs.io/en/latest/get_started/installation.html) and [MMSegmentation](https://github.com/open-mmlab/mmsegmentation/blob/master/docs/get_started.md#installation).\n```Shell\npip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/cu111/torch1.9.0/index.html\npip install mmsegmentation\n```\n\n## Data Generation\nThe training data is generated using ```leaderboard/team_code/auto_pilot.py```. Data generation requires routes and scenarios. Each route is defined by a sequence of waypoints (and optionally a weather condition) that the agent needs to follow. Each scenario is defined by a trigger transform (location and orientation) and other actors present in that scenario (optional). We provide several routes and scenarios under ```leaderboard/data/```. The [TransFuser repository](https://github.com/autonomousvision/transfuser) and [leaderboard repository](https://github.com/carla-simulator/leaderboard/tree/master/data) provide additional routes and scenario files.\n\n### Running a CARLA Server\n\n#### With Display\n```Shell\n./CarlaUE4.sh --world-port=2000 -opengl\n```\n\n#### Without Display\n\nWithout Docker:\n```Shell\nSDL_VIDEODRIVER=offscreen SDL_HINT_CUDA_DEVICE=0 ./CarlaUE4.sh --world-port=2000 -opengl\n```\n\nWith Docker:\n\nInstructions for setting up docker are available [here](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html#docker). Pull the docker image of CARLA 0.9.10.1 ```docker pull carlasim/carla:0.9.10.1```.\n\nDocker 18:\n```Shell\ndocker run -it --rm -p 2000-2002:2000-2002 --runtime=nvidia -e NVIDIA_VISIBLE_DEVICES=0 carlasim/carla:0.9.10.1 ./CarlaUE4.sh --world-port=2000 -opengl\n```\n\nDocker 19:\n```Shell\ndocker run -it --rm --net=host --gpus '\"device=0\"' carlasim/carla:0.9.10.1 ./CarlaUE4.sh --world-port=2000 -opengl\n```\n\nIf the docker container doesn't start properly then add another environment variable ```-e SDL_AUDIODRIVER=dsp```.\n\n### Running the Autopilot\n\nOnce the CARLA server is running, rollout the autopilot to start data generation.\n```Shell\n./leaderboard/scripts/run_evaluation.sh\n```\nThe expert agent used for data generation is defined in ```leaderboard/team_code/auto_pilot.py```. Different variables which need to be set are specified in ```leaderboard/scripts/run_evaluation.sh```. The expert agent is originally based on the autopilot from [this codebase](https://github.com/bradyz/2020_CARLA_challenge).\n\n## Training\nThe training code and pretrained models are provided below.\n```Shell\nmkdir model_ckpt\nwget https://s3.eu-central-1.amazonaws.com/avg-projects/neat/models.zip -P model_ckpt\nunzip model_ckpt/models.zip -d model_ckpt/\nrm model_ckpt/models.zip\n```\n\nThere are 5 pretrained models provided in ```model_ckpt/```:\n- [AIM-MT (2D)](leaderboard/team_code/aim_mt_2d_agent.py): ```aim_mt_sem``` and ```aim_mt_sem_depth```\n- [AIM-MT (BEV)](leaderboard/team_code/aim_mt_bev_agent.py): ```aim_mt_bev```\n- [AIM-VA](leaderboard/team_code/aim_va_agent.py): ```aim_va```\n- [NEAT](leaderboard/team_code/neat_agent.py): ```neat```\n\nAdditional baselines are available in the [TransFuser repository](https://github.com/autonomousvision/transfuser).\n\n## Evaluation\nSpin up a CARLA server (described above) and run the required agent. The required variables need to be set in ```leaderboard/scripts/run_evaluation.sh```.\n```Shell\nCUDA_VISIBLE_DEVICES=0 ./leaderboard/scripts/run_evaluation.sh\n```\n\n## Acknowledgements\nThis implementation primarily extends the [cvpr2021 branch of the existing TransFuser repository](https://github.com/autonomousvision/transfuser/tree/cvpr2021).\n\nIf you found our work interesting, check out the code for some more recent work on CARLA from our group:\n- [Renz et al., PlanT: Explainable Planning Transformers via Object-Level Representations (CoRL 2022)](https://github.com/autonomousvision/plant)\n- [Chitta et al., TransFuser: Imitation with Transformer-Based Sensor Fusion for Autonomous Driving (PAMI 2022)](https://github.com/autonomousvision/transfuser)\n- [Hanselmann et al., KING: Generating Safety-Critical Driving Scenarios for Robust Imitation via Kinematics Gradients (ECCV 2022)](https://github.com/autonomousvision/king)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fautonomousvision%2Fneat","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fautonomousvision%2Fneat","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fautonomousvision%2Fneat/lists"}