{"id":13441768,"url":"https://github.com/MCG-NJU/LinK","last_synced_at":"2025-03-20T12:32:46.180Z","repository":{"id":169826279,"uuid":"619011709","full_name":"MCG-NJU/LinK","owner":"MCG-NJU","description":"[CVPR 2023] LinK: Linear Kernel for LiDAR-based 3D Perception","archived":false,"fork":false,"pushed_at":"2024-07-27T13:57:22.000Z","size":535,"stargazers_count":85,"open_issues_count":3,"forks_count":5,"subscribers_count":7,"default_branch":"main","last_synced_at":"2024-10-28T04:19:30.195Z","etag":null,"topics":["autonomous-driving","detection","large-kernels","nuscenes","point-cloud","segmentation","semantickitti"],"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/MCG-NJU.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":"2023-03-26T01:54:10.000Z","updated_at":"2024-10-14T09:59:06.000Z","dependencies_parsed_at":null,"dependency_job_id":"438bc7ae-fc56-4743-bb69-f04cf4b341f2","html_url":"https://github.com/MCG-NJU/LinK","commit_stats":{"total_commits":8,"total_committers":2,"mean_commits":4.0,"dds":0.25,"last_synced_commit":"a91a9a5744062ae5e4f1d689ff3bf125897b4444"},"previous_names":["mcg-nju/link"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/MCG-NJU%2FLinK","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/MCG-NJU%2FLinK/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/MCG-NJU%2FLinK/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/MCG-NJU%2FLinK/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/MCG-NJU","download_url":"https://codeload.github.com/MCG-NJU/LinK/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":244611437,"owners_count":20481192,"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":["autonomous-driving","detection","large-kernels","nuscenes","point-cloud","segmentation","semantickitti"],"created_at":"2024-07-31T03:01:37.849Z","updated_at":"2025-03-20T12:32:46.174Z","avatar_url":"https://github.com/MCG-NJU.png","language":"Python","funding_links":[],"categories":["Python"],"sub_categories":[],"readme":"# [LinK: Linear Kernel for LiDAR-based 3D Perception](https://arxiv.org/abs/2303.16094)\n\nOfficial PyTorch implementation of **LinK**, from the following paper: \n\n[LinK: Linear Kernel for LiDAR-based 3D Perception](https://arxiv.org/abs/2303.16094). CVPR 2023.\\\n[Tao Lu](https://github.com/inspirelt), [Xiang Ding](https://github.com/shawnding), [Haisong Liu](https://github.com/afterthat97), Gangshan Wu, [Limin Wang](https://wanglimin.github.io/) \u003cbr /\u003e\nMultimedia Computing Group, Nanjing University \\\n[[`arxiv`](https://arxiv.org/abs/2303.16094)][[`Conference version`](https://openaccess.thecvf.com/content/CVPR2023/html/Lu_LinK_Linear_Kernel_for_LiDAR-Based_3D_Perception_CVPR_2023_paper.html)]\n\n\n--- \n\n\u003cp align=\"center\"\u003e\n\u003cimg src=\"figs/link.png\" width=100% height=100% \nclass=\"center\"\u003e\n\u003c/p\u003e\n\n\n**LinK** is a large kernel backbone for 3D perception tasks, consisting of a linear kernel generator and a pre-aggregation strategy. The two designs scale up the perception range into 21x21x21 with linear complexity.\n\n---\n\n## Model Zoo\n### Segmentation on SemanticKITTI(val)\n\n| name | kernel config |mIoU | model |\n|:---:|:---:|:---:|:---:|\n| LinK | cos_x:(2x3)^3 | 67.72 | [model](https://drive.google.com/file/d/1v5K2JlJ2QVmNA_D74SGxM_bvA_dRHO6d/view?usp=sharing) |\n| LinK | cos:(3x7)^3 | 67.50 | [model](https://drive.google.com/file/d/1mgo_SVwEhWXS3OaGGwxmYKOOXMWOkKRi/view?usp=sharing) |\n| LinK(encoder-only) | cos_x:(2x3)^3 | 67.33 | [model](https://drive.google.com/file/d/1TRsJOoL_H2v5VkXd9JMrklenw9Oy4yzQ/view?usp=sharing)|\n| LinK(encoder-only) | cos:(3x5)^3 | 67.07 | [model](https://drive.google.com/file/d/1tccURuCPFnaPHg_7s82PBbb4cJXF388A/view?usp=sharing) |\n\n\n\n### Detection on nuScenes\n\n- Validation\n\n| name | kernel config |NDS | mAP | model |\n|:---:|:---:|:---:|:---:|:---:|\n| LinK | cos:(3x7)^3 | 69.5 | 63.6 | [model](https://drive.google.com/file/d/11aGVrwkAno8baqh9YgPahM6y4iSSl5ik/view?usp=sharing) |\n\n- Test\n\n| name | kernel config |NDS | mAP | model |\n|:---:|:---:|:---:|:---:|:---:|\n| LinK | cos:(3x7)^3 | 71.0 | 66.3 | [model](https://drive.google.com/file/d/1I8hr6XLHNAIfL4LIeBn0Co87s_DEO6Oy/view?usp=sharing) |\n| LinK(TTA) | cos:(3x7)^3 | 73.4 | 69.8 | [model](https://drive.google.com/file/d/1I8hr6XLHNAIfL4LIeBn0Co87s_DEO6Oy/view?usp=sharing) |\n\n\n## Installation\n\nClone this repo to your workspace. \n```\ngit clone https://github.com/MCG-NJU/LinK.git\ncd LinK\n```\n\n## Semantic Segmentation\n\nplease check [segmentation/INSTALL.md](segmentation/INSTALL.md) and [segmentation/GET_STARTED.md](segmentation/GET_STARTED.md).\n\n## Detection\n\nsee [detection/INSTALL.md](detection/INSTALL.md) and [detection/GET_STARTED.md](detection/GET_STARTED.md).\n\n\n## Citation\n\nIf you find our work helpful, please consider citing:\n```\n@InProceedings{lu2023link,\n    author    = {Lu, Tao and Ding, Xiang and Liu, Haisong and Wu, Gangshan and Wang, Limin},\n    title     = {LinK: Linear Kernel for LiDAR-Based 3D Perception},\n    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},\n    month     = {June},\n    year      = {2023},\n    pages     = {1105-1115}\n}\n```\n\n```\n@article{lu2022app,\n  title={APP-Net: Auxiliary-point-based Push and Pull Operations for Efficient Point Cloud Classification},\n  author={Lu, Tao and Liu, Chunxu and Chen, Youxin and Wu, Gangshan and Wang, Limin},\n  journal={arXiv preprint arXiv:2205.00847},\n  year={2022}\n}\n```\n\n\n## Contact\n- Tao Lu: taolu@smail.nju.edu.cn\n- Xiang Ding: xding@smail.nju.edu.cn\n- Haisong Liu: liuhs@smail.nju.edu.cn\n\n## Acknowledgement\n\nOur code is based on [CenterPoint](https://github.com/tianweiy/CenterPoint), [SPVNAS](https://github.com/mit-han-lab/spvnas), [spconv](https://github.com/traveller59/spconv), [torchsparse](https://github.com/mit-han-lab/torchsparse). And we thank a lot for the kind help from [Ruixiang Zhang](https://github.com/zxczrx123), [Xu Yan](https://yanx27.github.io/) and [Yukang Chen](https://yukangchen.com/).\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FMCG-NJU%2FLinK","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FMCG-NJU%2FLinK","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FMCG-NJU%2FLinK/lists"}