{"id":13441933,"url":"https://github.com/Sense-X/HoP","last_synced_at":"2025-03-20T13:31:31.880Z","repository":{"id":181670696,"uuid":"666418871","full_name":"Sense-X/HoP","owner":"Sense-X","description":"[ICCV 2023] Temporal Enhanced Training of Multi-view 3D Object Detector via Historical Object Prediction","archived":false,"fork":false,"pushed_at":"2023-08-24T01:20:30.000Z","size":4010,"stargazers_count":179,"open_issues_count":10,"forks_count":13,"subscribers_count":8,"default_branch":"main","last_synced_at":"2024-10-28T05:12:13.805Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/Sense-X.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":".github/CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":".github/CODE_OF_CONDUCT.md","threat_model":null,"audit":null,"citation":"CITATION.cff","codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null}},"created_at":"2023-07-14T13:20:02.000Z","updated_at":"2024-09-28T15:37:09.000Z","dependencies_parsed_at":"2024-01-16T02:56:22.124Z","dependency_job_id":null,"html_url":"https://github.com/Sense-X/HoP","commit_stats":null,"previous_names":["sense-x/hop"],"tags_count":1,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Sense-X%2FHoP","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Sense-X%2FHoP/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Sense-X%2FHoP/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Sense-X%2FHoP/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Sense-X","download_url":"https://codeload.github.com/Sense-X/HoP/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":244619148,"owners_count":20482369,"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":[],"created_at":"2024-07-31T03:01:39.786Z","updated_at":"2025-03-20T13:31:31.875Z","avatar_url":"https://github.com/Sense-X.png","language":"Python","funding_links":[],"categories":["Python"],"sub_categories":[],"readme":"# Temporal Enhanced Training of Multi-view 3D Object Detector via Historical Object Prediction\n\n[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/temporal-enhanced-training-of-multi-view-3d/3d-object-detection-on-nuscenes-camera-only)](https://paperswithcode.com/sota/3d-object-detection-on-nuscenes-camera-only?p=temporal-enhanced-training-of-multi-view-3d)\n\nThis repo is the official implementation of [\"Temporal Enhanced Training of Multi-view 3D Object Detector via Historical Object Prediction\"](https://arxiv.org/abs/2304.00967) by Zhuofan Zong, Dongzhi Jiang, Guanglu Song, Zeyue Xue, Jingyong Su, Hongsheng Li, and Yu Liu.\n\n\n## News\n* ***[07/25/2023]*** Code for HoP on BEVDet is released!\n* ***[07/14/2023]*** HoP is accepted to ICCV 2023!\n* ***[04/05/2023]*** HoP achieves new SOTA performance on [nuScenes 3D detection leaderboard](https://www.nuscenes.org/object-detection?externalData=all\u0026mapData=all\u0026modalities=Camera) with **68.5 NDS** and **62.4 mAP**.\n\n## Model Zoo\n\n### Result on BEVDet4D-Depth\n\n|          model           | backbone |   pretrain   | img size | Epoch |  NDS   |  mAP   |                            config                            |                             ckpt                             |                             log                              |\n| :----------------------: | :------: | :----------: | :------: | :---: | :----: | :----: | :----------------------------------------------------------: | :----------------------------------------------------------: | :----------------------------------------------------------: |\n| BEVDet4D-Depth(Baseline) |  Res50   | [ImageNet]() | 256x704  |  24   | 0.4930 | 0.3848 | [cfg](https://github.com/Sense-X/HoP/blob/main/configs/hop_bevdet/bevdet4d-r50-depth.py) | [ckpt](https://github.com/Sense-X/HoP/releases/download/Release/BEVDet_ep24_ema.pth) | [log](https://github.com/Sense-X/HoP/releases/download/Release/BEVDet.log) |\n|    HoP_BEVDet4D-Depth    |  Res50   | [ImageNet]() | 256x704  |  24   | 0.5099 | 0.3990 | [cfg](https://github.com/Sense-X/HoP/blob/main/configs/hop_bevdet/hop_bevdet4d-r50-depth.py) | [ckpt](https://github.com/Sense-X/HoP/releases/download/Release/HoP_BEVDet_ep24_ema.pth) | [log](https://github.com/Sense-X/HoP/releases/download/Release/HoP_BEVDet.log) |\n\n## Get Started\n\n### Install\n\nWe train our models under the following environment: \n\n```\npython=3.6.9\npytorch=1.8.1\ntorchvision=0.9.1\ncuda=11.2\n```\n\nOther versions may possibly be imcompatible.\n\nWe use [MMDetection3D V1.0.0rc4](https://github.com/open-mmlab/mmdetection3d/tree/v1.0.0rc4), [MMDetection V2.24.0](https://github.com/open-mmlab/mmdetection/releases/tag/v2.25.3) and [MMCV V1.5.0](https://github.com/open-mmlab/mmcv/releases/tag/v1.5.0). The source code of MMDetection3D has been included in this repo.\n\nYou can take the following steps to install packages above: \n\n1. Build MMCV following [official instructions](https://github.com/open-mmlab/mmcv/tree/v1.5.2#installation). \n\n2. Install MMDetection by \n\n   ```bash\n   pip install mmdet==2.24.0\n   ```\n\n3. Copy HoP repo and install MMDetection3D.\n\n   ```bash\n   git clone git@github.com:Sense-X/HoP.git\n   cd HoP\n   pip install -e .\n   ```\n\n### Data Preparation\n\nFollow the steps to prepare nuScenes Dataset introduced in [nuscenes_det.md](https://github.com/HuangJunJie2017/BEVDet/blob/dev2.1/docs/en/datasets/nuscenes_det.md) and create the pkl by running:\n\n```bash\npython tools/create_data_bevdet.py\n```\n\n### Train HoP\n\n```bash\n# single gpu\npython tools/train.py configs/hop_bevdet/hop_bevdet4d-r50-depth.py\n# multiple gpu\n./tools/dist_train.sh configs/hop_bevdet/hop_bevdet4d-r50-depth.py $num_gpu\n```\n\n### Eval HoP\n\n```bash\n# single gpu\npython tools/test.py configs/hop_bevdet/hop_bevdet4d-r50-depth.py $checkpoint --eval bbox\n# multiple gpu\n./tools/dist_test.sh configs/hop_bevdet/hop_bevdet4d-r50-depth.py $checkpoint $num_gpu --eval bbox\n```\n\n## Method\n\n\u003cimg src=\"resources/HoP_framework.png\" width=\"1000\" \u003e\n\n## TODO\n\n- [ ] Release code for HoP on BEVFormer.\n\n\n## Cite HoP\n\nIf you find this repository useful, please use the following BibTeX entry for citation.\n\n```latex\n@misc{hop2023,\n      title={Temporal Enhanced Training of Multi-view 3D Object Detector via Historical Object Prediction},\n      author={Zhuofan Zong and Dongzhi Jiang and Guanglu Song and Zeyue Xue and Jingyong Su and Hongsheng Li and Yu Liu},\n      year={2023},\n      eprint={2304.00967},\n      archivePrefix={arXiv},\n      primaryClass={cs.CV}\n}\n```\n\n## License\n\nThis project is released under the MIT license. Please see the [LICENSE](LICENSE) file for more information.","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FSense-X%2FHoP","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FSense-X%2FHoP","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FSense-X%2FHoP/lists"}