{"id":27643330,"url":"https://github.com/matrixorigin/matrix-light-reid","last_synced_at":"2026-04-17T02:31:39.521Z","repository":{"id":102992425,"uuid":"355920343","full_name":"matrixorigin/matrix-light-reid","owner":"matrixorigin","description":null,"archived":false,"fork":false,"pushed_at":"2022-10-13T18:43:12.000Z","size":1745,"stargazers_count":1,"open_issues_count":0,"forks_count":1,"subscribers_count":4,"default_branch":"main","last_synced_at":"2025-08-24T00:39:50.544Z","etag":null,"topics":["computer-vision","person-reid","person-reidentification","pytorch","re-identification"],"latest_commit_sha":null,"homepage":"","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/matrixorigin.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,"zenodo":null}},"created_at":"2021-04-08T13:34:14.000Z","updated_at":"2022-06-23T08:27:52.000Z","dependencies_parsed_at":"2023-03-13T15:10:06.983Z","dependency_job_id":null,"html_url":"https://github.com/matrixorigin/matrix-light-reid","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/matrixorigin/matrix-light-reid","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/matrixorigin%2Fmatrix-light-reid","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/matrixorigin%2Fmatrix-light-reid/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/matrixorigin%2Fmatrix-light-reid/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/matrixorigin%2Fmatrix-light-reid/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/matrixorigin","download_url":"https://codeload.github.com/matrixorigin/matrix-light-reid/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/matrixorigin%2Fmatrix-light-reid/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":31912344,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-16T18:22:33.417Z","status":"online","status_checked_at":"2026-04-17T02:00:06.879Z","response_time":62,"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":["computer-vision","person-reid","person-reidentification","pytorch","re-identification"],"created_at":"2025-04-24T00:12:58.581Z","updated_at":"2026-04-17T02:31:39.508Z","avatar_url":"https://github.com/matrixorigin.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"\n## Matrix Light ReID (MLR)\n\nMLR is a Person-ReID training framework based on [mmcls](https://github.com/open-mmlab/mmclassification), which supports both ResNet family backbones and mobile level backbones, with modular design.\n\nWe also support network searching to find best accuracy-latency tradeoff, for lightweight models.\n\n\n## Baseline Performance\n\n| Backbone | Head        | mAP   | Rank-1 | FLOPs |\n|----------|-------------|-------|--------|-------|\n| ResNet50 | BagOfTricks | 85.53 | 93.91  | 4.08G |\n|          |             |       |        |       |\n\n\n[more]\n\n## Data Prepare\n\nUnzip Market1501 dataset to ``data/Market-1501-v15.09.15/``\n\n\n## Training\n\n\nFor training bag-of-tricks model:\n\n``CUDA_VISIBLE_DEVICES='0' PORT=29711 bash tools/dist_train.sh ./configs/reid/resnet50_market1501.py 1``\n\n\n## Evaluation\n\nAs default, the training process will evaluate the scores of mAP and Rank-1 every 10 epochs.\n\n\n## Network Search for Lightweight backbones\n\nAs for the name **Light** ReID, we try to find a lightweight solution to produce high quality ReID features.\n\nHere we use the MobileNetV2 as backbone and the same optimizer from ResNet50 baseline as our base solution. Our aim is to find a better backbone and training strategy to obtain good ReID mAP and Rank-1 accuracy, under the train-from-scratch situation.\n\n| Backbone     | feat-dim | Optimizer               | LR epoch steps  | mAP       | Rank-1    | FLOPs |\n| ------------ | -------- | ----------------------- | --------------- | --------- | --------- | ----- |\n| MobileNetV2  | 256      | Adam, wd 5e-4           | 50,120,150      | 46.96     | 71.17     | 210M  |\n| MobileNetV2  | 256      | SGD, wd 1e-4            | 50,120,150      | 47.70     | 71.61     | 210M  |\n| MobileNetV2  | 256      | SGD, wd 5e-4            | 50,120,150      | 61.62     | 81.32     | 210M  |\n| MLR-210M-A   | 256      | SGD, wd 5e-4            | 50,120,150      | 64.57     | 84.56     | 212M  |\n| MLR-210M-B   | 256      | SGD, wd 5e-4            | 50,120,150      | 65.23     | 83.88     | 207M  |\n| MLR-210M-A+  | 256      | SGD, wd 5e-4, Warmup-2K | 150,250,320,350 | 69.18     | 86.13     | 212M  |\n| MLR-210M-B+  | 256      | SGD, wd 5e-4, Warmup-2K | 150,250,320,350 | **71.20** | **87.77** | 207M  |\n| MLR-210M-B++ | 512      | SGD, wd 5e-4, Warmup-2K | 150,250,320,350 | 71.19     | 87.35     | 207M  |\n\n## Reference\n\nSome code is inspired by [fastreid](https://github.com/JDAI-CV/fast-reid) and [gluon-reid](https://github.com/xiaolai-sqlai/gluon-reid)\n\n\nRef-Papers:\n\n```\n@inproceedings{luo2019bag,\n  title={Bag of tricks and a strong baseline for deep person re-identification},\n  author={Luo, Hao and Gu, Youzhi and Liao, Xingyu and Lai, Shenqi and Jiang, Wei},\n  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops},\n  pages={0--0},\n  year={2019}\n}\n```\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmatrixorigin%2Fmatrix-light-reid","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fmatrixorigin%2Fmatrix-light-reid","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmatrixorigin%2Fmatrix-light-reid/lists"}