{"id":13441702,"url":"https://github.com/SamsungLabs/tr3d","last_synced_at":"2025-03-20T12:32:30.050Z","repository":{"id":108876267,"uuid":"596000075","full_name":"SamsungLabs/tr3d","owner":"SamsungLabs","description":"[ICIP2023] TR3D: Towards Real-Time Indoor 3D Object Detection","archived":false,"fork":false,"pushed_at":"2023-09-25T15:43:45.000Z","size":9960,"stargazers_count":136,"open_issues_count":12,"forks_count":8,"subscribers_count":7,"default_branch":"main","last_synced_at":"2024-08-01T03:37:18.564Z","etag":null,"topics":["3d-object-detection","pytorch","s3dis","scannet","sunrgbd","tr3d"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/SamsungLabs.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}},"created_at":"2023-02-01T08:54:04.000Z","updated_at":"2024-07-22T02:50:32.000Z","dependencies_parsed_at":"2023-03-19T01:03:36.754Z","dependency_job_id":"aa60ff73-96d2-4e49-a49b-343ccf6074b5","html_url":"https://github.com/SamsungLabs/tr3d","commit_stats":null,"previous_names":[],"tags_count":1,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/SamsungLabs%2Ftr3d","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/SamsungLabs%2Ftr3d/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/SamsungLabs%2Ftr3d/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/SamsungLabs%2Ftr3d/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/SamsungLabs","download_url":"https://codeload.github.com/SamsungLabs/tr3d/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":221760123,"owners_count":16876357,"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","pytorch","s3dis","scannet","sunrgbd","tr3d"],"created_at":"2024-07-31T03:01:37.164Z","updated_at":"2025-03-20T12:32:30.017Z","avatar_url":"https://github.com/SamsungLabs.png","language":"Python","funding_links":[],"categories":["Python"],"sub_categories":[],"readme":"[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/tr3d-towards-real-time-indoor-3d-object/3d-object-detection-on-scannetv2)](https://paperswithcode.com/sota/3d-object-detection-on-scannetv2?p=tr3d-towards-real-time-indoor-3d-object)\n[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/tr3d-towards-real-time-indoor-3d-object/3d-object-detection-on-sun-rgbd-val)](https://paperswithcode.com/sota/3d-object-detection-on-sun-rgbd-val?p=tr3d-towards-real-time-indoor-3d-object)\n[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/tr3d-towards-real-time-indoor-3d-object/3d-object-detection-on-s3dis)](https://paperswithcode.com/sota/3d-object-detection-on-s3dis?p=tr3d-towards-real-time-indoor-3d-object)\n\n## TR3D: Towards Real-Time Indoor 3D Object Detection\n\n**News**:\n * :fire: June, 2023. TR3D is accepted at [ICIP2023](https://2023.ieeeicip.org/).\n * :rocket: June, 2023. We add ScanNet-pretrained S3DIS model and log significantly pushing forward state-of-the-art.\n * February, 2023. TR3D on all 3 datasets is now supported in [mmdetection3d](https://github.com/open-mmlab/mmdetection3d) as a [project](https://github.com/open-mmlab/mmdetection3d/tree/main/projects/TR3D).\n * :fire: February, 2023. TR3D is now state-of-the-art on [paperswithcode](https://paperswithcode.com) on SUN RGB-D and S3DIS.\n\nThis repository contains an implementation of TR3D, a 3D object detection method introduced in our paper:\n\n\u003e **TR3D: Towards Real-Time Indoor 3D Object Detection**\u003cbr\u003e\n\u003e [Danila Rukhovich](https://github.com/filaPro),\n\u003e [Anna Vorontsova](https://github.com/highrut),\n\u003e [Anton Konushin](https://scholar.google.com/citations?user=ZT_k-wMAAAAJ)\n\u003e \u003cbr\u003e\n\u003e Samsung Research\u003cbr\u003e\n\u003e https://arxiv.org/abs/2302.02858\n\n### Installation\nFor convenience, we provide a [Dockerfile](docker/Dockerfile).\n\nAlternatively, you can install all required packages manually. This implementation is based on [mmdetection3d](https://github.com/open-mmlab/mmdetection3d) framework.\nPlease refer to the original installation guide [getting_started.md](docs/en/getting_started.md), including MinkowskiEngine installation, replacing `open-mmlab/mmdetection3d` with `samsunglabs/tr3d`.\n\n\nMost of the `TR3D`-related code locates in the following files: \n[detectors/mink_single_stage.py](mmdet3d/models/detectors/mink_single_stage.py),\n[detectors/tr3d_ff.py](mmdet3d/models/detectors/tr3d_ff.py),\n[dense_heads/tr3d_head.py](mmdet3d/models/dense_heads/tr3d_head.py),\n[necks/tr3d_neck.py](mmdet3d/models/necks/tr3d_neck.py).\n\n### Getting Started\n\nPlease see [getting_started.md](docs/getting_started.md) for basic usage examples.\nWe follow the mmdetection3d data preparation protocol described in [scannet](data/scannet), [sunrgbd](data/sunrgbd), and [s3dis](data/s3dis).\n\n**Training**\n\nTo start training, run [train](tools/train.py) with TR3D [configs](configs/tr3d):\n```shell\npython tools/train.py configs/tr3d/tr3d_scannet-3d-18class.py\n```\n\n**Testing**\n\nTest pre-trained model using [test](tools/dist_test.sh) with TR3D [configs](configs/tr3d):\n```shell\npython tools/test.py configs/tr3d/tr3d_scannet-3d-18class.py \\\n    work_dirs/tr3d_scannet-3d-18class/latest.pth --eval mAP\n```\n\n**Visualization**\n\nVisualizations can be created with [test](tools/test.py) script. \nFor better visualizations, you may set `score_thr` in configs to `0.3`:\n```shell\npython tools/test.py configs/tr3d/tr3d_scannet-3d-18class.py \\\n    work_dirs/tr3d_scannet-3d-18class/latest.pth --eval mAP --show \\\n    --show-dir work_dirs/tr3d_scannet-3d-18class\n```\n\n### Models\n\nThe metrics are obtained in 5 training runs followed by 5 test runs. We report both the best and the average values (the latter are given in round brackets).\nInference speed (scenes per second) is measured on a single NVidia RTX 4090. Please, note that ScanNet-pretrained S3DIS model was actually trained in the original\n[openmmlab/mmdetection3d](https://github.com/open-mmlab/mmdetection3d/tree/main/projects/TR3D) codebase.\n\n**TR3D 3D Detection**\n\n| Dataset | mAP@0.25 | mAP@0.5 | Scenes \u003cbr\u003e per sec.| Download |\n|:-------:|:--------:|:-------:|:-------------------:|:--------:|\n| ScanNet | 72.9 (72.0) | 59.3 (57.4) | 23.7 | [model](https://github.com/samsunglabs/tr3d/releases/download/v1.0/tr3d_scannet.pth) \u0026#124; [log](https://github.com/samsunglabs/tr3d/releases/download/v1.0/tr3d_scannet.log.json) \u0026#124; [config](configs/tr3d/tr3d_scannet-3d-18class.py) |\n| SUN RGB-D | 67.1 (66.3) | 50.4 (49.6) | 27.5 | [model](https://github.com/samsunglabs/tr3d/releases/download/v1.0/tr3d_sunrgbd.pth) \u0026#124; [log](https://github.com/samsunglabs/tr3d/releases/download/v1.0/tr3d_sunrgbd.log.json) \u0026#124; [config](configs/tr3d/tr3d_sunrgbd-3d-10class.py) |\n| S3DIS | 74.5 (72.1) | 51.7 (47.6) | 21.0 | [model](https://github.com/samsunglabs/tr3d/releases/download/v1.0/tr3d_s3dis.pth) \u0026#124; [log](https://github.com/samsunglabs/tr3d/releases/download/v1.0/tr3d_s3dis.log.json) \u0026#124; [config](configs/tr3d/tr3d_s3dis-3d-5class.py) |\n| S3DIS \u003cbr\u003e ScanNet-pretrained | 75.9 (75.1) | 56.6 (54.8) | 21.0 | [model](https://github.com/samsunglabs/tr3d/releases/download/v1.0/tr3d_scannet-pretrain_s3dis.pth) \u0026#124; [log](https://github.com/samsunglabs/tr3d/releases/download/v1.0/tr3d_scannet-pretrain_s3dis.log) \u0026#124; [config](configs/tr3d/tr3d_scannet-pretrain_s3dis-3d-5class.py) |\n\n**RGB + PC 3D Detection on SUN RGB-D**\n\n| Model | mAP@0.25 | mAP@0.5 | Scenes \u003cbr\u003e per sec.| Download |\n|:-----:|:--------:|:-------:|:-------------------:|:--------:|\n| ImVoteNet | 63.4 | - | 14.8 | [instruction](configs/imvotenet) |\n| VoteNet+FF | 64.5 (63.7) | 39.2 (38.1) | - | [model](https://github.com/samsunglabs/tr3d/releases/download/v1.0/votenet_ff_sunrgbd.pth) \u0026#124; [log](https://github.com/samsunglabs/tr3d/releases/download/v1.0/votenet_ff_sunrgbd.log.json) \u0026#124; [config](configs/votenet/votenet-ff_16x8_sunrgbd-3d-10class.py) |\n| TR3D+FF | 69.4 (68.7) | 53.4 (52.4) | 17.5 | [model](https://github.com/samsunglabs/tr3d/releases/download/v1.0/tr3d_ff_sunrgbd.pth) \u0026#124; [log](https://github.com/samsunglabs/tr3d/releases/download/v1.0/tr3d_ff_sunrgbd.log.json) \u0026#124; [config](configs/tr3d/tr3d-ff_sunrgbd-3d-10class.py) |\n\n### Example Detections\n\n\u003cp align=\"center\"\u003e\u003cimg src=\"./resources/github.png\" alt=\"drawing\" width=\"90%\"/\u003e\u003c/p\u003e\n\n### Citation\n\nIf you find this work useful for your research, please cite our paper:\n\n```\n@misc{rukhovich2023tr3d,\n  doi = {10.48550/ARXIV.2302.02858},\n  url = {https://arxiv.org/abs/2302.02858},\n  author = {Rukhovich, Danila and Vorontsova, Anna and Konushin, Anton},\n  title = {TR3D: Towards Real-Time Indoor 3D Object Detection},\n  publisher = {arXiv},\n  year = {2023}\n}\n```\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FSamsungLabs%2Ftr3d","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FSamsungLabs%2Ftr3d","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FSamsungLabs%2Ftr3d/lists"}