{"id":13443888,"url":"https://github.com/maudzung/SFA3D","last_synced_at":"2025-03-20T17:32:06.228Z","repository":{"id":37400434,"uuid":"289947862","full_name":"maudzung/SFA3D","owner":"maudzung","description":"Super Fast and Accurate 3D Object Detection based on 3D LiDAR Point Clouds (The PyTorch implementation)","archived":false,"fork":false,"pushed_at":"2023-09-17T01:57:42.000Z","size":46579,"stargazers_count":951,"open_issues_count":22,"forks_count":269,"subscribers_count":32,"default_branch":"master","last_synced_at":"2024-04-13T18:37:34.542Z","etag":null,"topics":["3d-object-detection","bevmap","center","fast-detection","kitti-dataset","lidar-point-cloud","real-time","ros","rtm3d"],"latest_commit_sha":null,"homepage":"https://github.com/maudzung/Super-Fast-Accurate-3D-Object-Detection","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/maudzung.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,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2020-08-24T14:16:33.000Z","updated_at":"2024-05-08T12:46:31.584Z","dependencies_parsed_at":"2024-05-08T12:46:29.090Z","dependency_job_id":"888c83df-d8b7-489f-a88b-45e39403fb6d","html_url":"https://github.com/maudzung/SFA3D","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/maudzung%2FSFA3D","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/maudzung%2FSFA3D/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/maudzung%2FSFA3D/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/maudzung%2FSFA3D/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/maudzung","download_url":"https://codeload.github.com/maudzung/SFA3D/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":244660628,"owners_count":20489367,"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","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":["3d-object-detection","bevmap","center","fast-detection","kitti-dataset","lidar-point-cloud","real-time","ros","rtm3d"],"created_at":"2024-07-31T03:02:12.980Z","updated_at":"2025-03-20T17:32:06.223Z","avatar_url":"https://github.com/maudzung.png","language":"Python","funding_links":[],"categories":["Python","3. Perception","Topics"],"sub_categories":["3.1.2 Lidar based","Perception"],"readme":"# Super Fast and Accurate 3D Object Detection based on 3D LiDAR Point Clouds (SFA3D)\n\n[![python-image]][python-url]\n[![pytorch-image]][pytorch-url]\n\n---\n\n## Features\n- [x] Super fast and accurate 3D object detection based on LiDAR\n- [x] Fast training, fast inference\n- [x] An Anchor-free approach\n- [x] No Non-Max-Suppression\n- [x] Support [distributed data parallel training](https://github.com/pytorch/examples/tree/master/distributed/ddp)\n- [x] Release pre-trained models \n\n## Highlights\n- [x] The technical details are described **[here](./Technical_details.md)**\n- [x] The great introduction and explanation from _`Computer Vision and Perception for Self-Driving Cars Course`_ **[Youtube link](https://youtu.be/cPOtULagNnI?t=4858)**\n- [x] SFA3D is used for the second course in the _`Udacity Self-Driving Car Engineer Nanodegree Program: Sensor Fusion and Tracking`_ **[GitHub link](https://github.com/udacity/nd013-c2-fusion-starter/tree/b1455b8ff433cb7f537d62e526209738293e7d8b)**\n\n**Update 2020.09.06**: Add `ROS` source code. The great work has been done by [@AhmedARadwan](https://github.com/AhmedARadwan). \nThe implementation is [here](https://github.com/maudzung/SFA3D/tree/ea0222c1b35489dc35d8452c989c4b014e20e0da)\n\n## Demonstration (on a single GTX 1080Ti)\n\n[![demo](http://img.youtube.com/vi/FI8mJIXkgX4/0.jpg)](http://www.youtube.com/watch?v=FI8mJIXkgX4)\n\n\n**[Youtube link](https://youtu.be/FI8mJIXkgX4)**\n\n## 2. Getting Started\n### 2.1. Requirement\n\nThe instructions for setting up a virtual environment is [here](https://github.com/maudzung/virtual_environment_python3).\n\n```shell script\ngit clone https://github.com/maudzung/SFA3D.git SFA3D\ncd SFA3D/\npip install -r requirements.txt\n```\n\n### 2.2. Data Preparation\nDownload the 3D KITTI detection dataset from [here](http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d).\n\nThe downloaded data includes:\n\n- Velodyne point clouds _**(29 GB)**_\n- Training labels of object data set _**(5 MB)**_\n- Camera calibration matrices of object data set _**(16 MB)**_\n- **Left color images** of object data set _**(12 GB)**_ (For visualization purpose only)\n\n\nPlease make sure that you construct the source code \u0026 dataset directories structure as below.\n\n### 2.3. How to run\n\n#### 2.3.1. Visualize the dataset \n\nTo visualize 3D point clouds with 3D boxes, let's execute:\n\n```shell script\ncd sfa/data_process/\npython kitti_dataset.py\n```\n\n\n#### 2.3.2. Inference\n\nThe pre-trained model was pushed to this repo.\n\n```\npython test.py --gpu_idx 0 --peak_thresh 0.2\n```\n\n#### 2.3.3. Making demonstration\n\n```\npython demo_2_sides.py --gpu_idx 0 --peak_thresh 0.2\n```\n\nThe data for the demonstration will be automatically downloaded by executing the above command.\n\n\n#### 2.3.4. Training\n\n##### 2.3.4.1. Single machine, single gpu\n\n```shell script\npython train.py --gpu_idx 0\n```\n\n##### 2.3.4.2. Distributed Data Parallel Training\n- **Single machine (node), multiple GPUs**\n\n```\npython train.py --multiprocessing-distributed --world-size 1 --rank 0 --batch_size 64 --num_workers 8\n```\n\n- **Two machines (two nodes), multiple GPUs**\n\n   - _**First machine**_\n    ```\n    python train.py --dist-url 'tcp://IP_OF_NODE1:FREEPORT' --multiprocessing-distributed --world-size 2 --rank 0 --batch_size 64 --num_workers 8\n    ```\n\n   - _**Second machine**_\n    ```\n    python train.py --dist-url 'tcp://IP_OF_NODE2:FREEPORT' --multiprocessing-distributed --world-size 2 --rank 1 --batch_size 64 --num_workers 8\n    ```\n\n#### Tensorboard\n\n- To track the training progress, go to the `logs/` folder and \n\n```shell script\ncd logs/\u003csaved_fn\u003e/tensorboard/\ntensorboard --logdir=./\n```\n\n- Then go to [http://localhost:6006/](http://localhost:6006/)\n\n\n## Contact\n\nIf you think this work is useful, please give me a star! \u003cbr\u003e\nIf you find any errors or have any suggestions, please contact me (**Email:** `nguyenmaudung93.kstn@gmail.com`). \u003cbr\u003e\nThank you!\n\n\n## Citation\n\n```bash\n@misc{Super-Fast-Accurate-3D-Object-Detection-PyTorch,\n  author =       {Nguyen Mau Dung},\n  title =        {{Super-Fast-Accurate-3D-Object-Detection-PyTorch}},\n  howpublished = {\\url{https://github.com/maudzung/Super-Fast-Accurate-3D-Object-Detection}},\n  year =         {2020}\n}\n```\n\n## References\n\n[1] CenterNet: [Objects as Points paper](https://arxiv.org/abs/1904.07850), [PyTorch Implementation](https://github.com/xingyizhou/CenterNet) \u003cbr\u003e\n[2] RTM3D: [PyTorch Implementation](https://github.com/maudzung/RTM3D) \u003cbr\u003e\n[3] Libra_R-CNN: [PyTorch Implementation](https://github.com/OceanPang/Libra_R-CNN)\n\n_The YOLO-based models with the same BEV maps input:_ \u003cbr\u003e\n[4] Complex-YOLO: [v4](https://github.com/maudzung/Complex-YOLOv4-Pytorch), [v3](https://github.com/ghimiredhikura/Complex-YOLOv3), [v2](https://github.com/AI-liu/Complex-YOLO)\n\n*3D LiDAR Point pre-processing:* \u003cbr\u003e\n[5] VoxelNet: [PyTorch Implementation](https://github.com/skyhehe123/VoxelNet-pytorch)\n\n## Folder structure\n\n```\n${ROOT}\n└── checkpoints/\n    ├── fpn_resnet_18/    \n        ├── fpn_resnet_18_epoch_300.pth\n└── dataset/    \n    └── kitti/\n        ├──ImageSets/\n        │   ├── test.txt\n        │   ├── train.txt\n        │   └── val.txt\n        ├── training/\n        │   ├── image_2/ (left color camera)\n        │   ├── calib/\n        │   ├── label_2/\n        │   └── velodyne/\n        └── testing/  \n        │   ├── image_2/ (left color camera)\n        │   ├── calib/\n        │   └── velodyne/\n        └── classes_names.txt\n└── sfa/\n    ├── config/\n    │   ├── train_config.py\n    │   └── kitti_config.py\n    ├── data_process/\n    │   ├── kitti_dataloader.py\n    │   ├── kitti_dataset.py\n    │   └── kitti_data_utils.py\n    ├── models/\n    │   ├── fpn_resnet.py\n    │   ├── resnet.py\n    │   └── model_utils.py\n    └── utils/\n    │   ├── demo_utils.py\n    │   ├── evaluation_utils.py\n    │   ├── logger.py\n    │   ├── misc.py\n    │   ├── torch_utils.py\n    │   ├── train_utils.py\n    │   └── visualization_utils.py\n    ├── demo_2_sides.py\n    ├── demo_front.py\n    ├── test.py\n    └── train.py\n├── README.md \n└── requirements.txt\n```\n\n\n\n[python-image]: https://img.shields.io/badge/Python-3.6-ff69b4.svg\n[python-url]: https://www.python.org/\n[pytorch-image]: https://img.shields.io/badge/PyTorch-1.5-2BAF2B.svg\n[pytorch-url]: https://pytorch.org/\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmaudzung%2FSFA3D","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fmaudzung%2FSFA3D","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmaudzung%2FSFA3D/lists"}