{"id":27942515,"url":"https://github.com/coincheung/selfsup","last_synced_at":"2025-07-13T17:40:16.463Z","repository":{"id":112693233,"uuid":"291669802","full_name":"CoinCheung/SelfSup","owner":"CoinCheung","description":"ssl method pretrain experiments and weights: mocov2 + fast-moco + regioncl + mixup + densecl","archived":false,"fork":false,"pushed_at":"2023-03-10T02:49:11.000Z","size":155,"stargazers_count":5,"open_issues_count":0,"forks_count":1,"subscribers_count":1,"default_branch":"master","last_synced_at":"2025-05-07T11:56:53.692Z","etag":null,"topics":["densecl","fast-moco","mixup","moco","pretraining","regioncl","self-supervised-learning"],"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/CoinCheung.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":"2020-08-31T09:18:02.000Z","updated_at":"2024-01-03T01:43:26.000Z","dependencies_parsed_at":null,"dependency_job_id":"a3ca2c5a-0b01-410f-9eaa-89a34e75a5e0","html_url":"https://github.com/CoinCheung/SelfSup","commit_stats":null,"previous_names":[],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/CoinCheung/SelfSup","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/CoinCheung%2FSelfSup","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/CoinCheung%2FSelfSup/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/CoinCheung%2FSelfSup/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/CoinCheung%2FSelfSup/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/CoinCheung","download_url":"https://codeload.github.com/CoinCheung/SelfSup/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/CoinCheung%2FSelfSup/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":265180562,"owners_count":23723712,"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":["densecl","fast-moco","mixup","moco","pretraining","regioncl","self-supervised-learning"],"created_at":"2025-05-07T11:56:51.732Z","updated_at":"2025-07-13T17:40:16.455Z","avatar_url":"https://github.com/CoinCheung.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"\n## experiment results\n\nEach model is train for 200 epoch.  \n\n\u003ctable\u003e\u003ctbody\u003e\n\u003c!-- START TABLE --\u003e\n\u003c!-- TABLE HEADER --\u003e\n\u003cth valign=\"bottom\"\u003e\u003c/th\u003e\n\u003cth valign=\"bottom\"\u003eIN-linear\u003c/th\u003e\n\u003cth valign=\"bottom\"\u003eIN-finetune\u003c/th\u003e\n\u003cth valign=\"bottom\"\u003ecoco-bbox\u003c/th\u003e\n\u003cth valign=\"bottom\"\u003ecoco-segm\u003c/th\u003e\n\u003cth valign=\"bottom\"\u003ecityscapes\u003c/th\u003e\n\u003cth valign=\"bottom\"\u003elink\u003c/th\u003e\n\u003c!-- TABLE BODY --\u003e\n\n\u003ctr\u003e\u003ctd align=\"left\"\u003e\u003ca href=\"https://arxiv.org/abs/2003.04297\"\u003emocov2 r50\u003c/a\u003e\u003c/td\u003e\n\u003ctd align=\"center\"\u003e67.36\u003c/td\u003e\n\u003ctd align=\"center\"\u003e77.07\u003c/td\u003e\n\u003ctd align=\"center\"\u003e38.68\u003c/td\u003e\n\u003ctd align=\"center\"\u003e33.88\u003c/td\u003e\n\u003ctd align=\"center\"\u003e77.88\u003c/td\u003e\n\u003ctd align=\"center\"\u003e\u003ca href=\"https://github.com/CoinCheung/SelfSup/releases/download/0.0.0/selfsup-model_1.tar\"\u003elink\u003c/a\u003e\u003c/td\u003e\n\u003c/tr\u003e\n\n\u003ctr\u003e\u003ctd align=\"left\"\u003e\u003ca href=\"https://arxiv.org/abs/2207.08220\"\u003e+fast-moco\u003c/a\u003e\u003c/td\u003e\n\u003ctd align=\"center\"\u003e70.83\u003c/td\u003e\n\u003ctd align=\"center\"\u003e77.16\u003c/td\u003e\n\u003ctd align=\"center\"\u003e39.30\u003c/td\u003e\n\u003ctd align=\"center\"\u003e34.38\u003c/td\u003e\n\u003ctd align=\"center\"\u003e77.94\u003c/td\u003e\n\u003ctd align=\"center\"\u003e\u003ca href=\"https://github.com/CoinCheung/SelfSup/releases/download/0.0.0/selfsup-model_2.tar\"\u003elink\u003c/a\u003e\u003c/td\u003e\n\u003c/tr\u003e\n\n\u003ctr\u003e\u003ctd align=\"left\"\u003e\u003ca href=\"https://arxiv.org/abs/2111.12309\"\u003e+cutmix\u003c/a\u003e\u003c/td\u003e\n\u003ctd align=\"center\"\u003e71.32\u003c/td\u003e\n\u003ctd align=\"center\"\u003e77.15\u003c/td\u003e\n\u003ctd align=\"center\"\u003e39.41\u003c/td\u003e\n\u003ctd align=\"center\"\u003e34.47\u003c/td\u003e\n\u003ctd align=\"center\"\u003e78.63\u003c/td\u003e\n\u003ctd align=\"center\"\u003e\u003ca href=\"https://github.com/CoinCheung/SelfSup/releases/download/0.0.0/r50_checkpoint_0199_mocov2_fastm_cutmix.pth.tar\"\u003elink\u003c/a\u003e\u003c/td\u003e\n\u003c/tr\u003e\n\n\u003ctr\u003e\u003ctd align=\"left\"\u003e\u003ca href=\"https://arxiv.org/abs/1710.09412\"\u003e+mixup\u003c/a\u003e\u003c/td\u003e\n\u003ctd align=\"center\"\u003e70.42\u003c/td\u003e\n\u003ctd align=\"center\"\u003e77.28\u003c/td\u003e\n\u003ctd align=\"center\"\u003e39.46\u003c/td\u003e\n\u003ctd align=\"center\"\u003e34.56\u003c/td\u003e\n\u003ctd align=\"center\"\u003e78.54\u003c/td\u003e\n\u003ctd align=\"center\"\u003e\u003ca href=\"https://github.com/CoinCheung/SelfSup/releases/download/0.0.0/r50_checkpoint_0199_mocov2_fastm_cutmix_mixup.pth.tar\"\u003elink\u003c/a\u003e\u003c/td\u003e\n\u003c/tr\u003e\n\n\u003ctr\u003e\u003ctd align=\"left\"\u003e\u003ca href=\"https://arxiv.org/abs/2011.09157\"\u003e+dense\u003c/a\u003e\u003c/td\u003e\n\u003ctd align=\"center\"\u003e68.79\u003c/td\u003e\n\u003ctd align=\"center\"\u003e77.28\u003c/td\u003e\n\u003ctd align=\"center\"\u003e40.00\u003c/td\u003e\n\u003ctd align=\"center\"\u003e34.81\u003c/td\u003e\n\u003ctd align=\"center\"\u003e78.69\u003c/td\u003e\n\u003ctd align=\"center\"\u003e\u003ca href=\"https://github.com/CoinCheung/SelfSup/releases/download/0.0.0/r50_checkpoint_0199_mocov2_fastm_cutmix_mixup_dense.pth.tar\"\u003elink\u003c/a\u003e\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\u003c/table\u003e\n\n\nNotes:   \n\u0026#8195;\u0026#8195;**IN-linear:**  linear evaluation on imagenet.   \n\u0026#8195;\u0026#8195;**IN-finetune:**  finetune on imagenet.   \n\u0026#8195;\u0026#8195;**coco-bbox:**  object detection on coco.   \n\u0026#8195;\u0026#8195;**coco-segm:**  instance segmentation on coco.  \n\u0026#8195;\u0026#8195;**cityscapes:**  semantic segmentation on cityscapes.   \n\u0026nbsp;\n\n\n## training platform: \n\n* ubuntu 18.04\n* 32 nvidia Tesla T4 gpu, driver 450.80.02\n* cuda 11.3\n* cudnn 8\n* miniconda python 3.8.8\n* pytorch 1.12.0\n\n\n\n\n## raw results\nEach experiment is done 4 times, and above result in the table is the mean of the 4 results.  \n\n\nmocov2:   \n\u0026#8195;linear:  \n\u0026#8195;\u0026#8195;Acc@1 67.416 Acc@5 87.872  \n\u0026#8195;\u0026#8195;Acc@1 67.312 Acc@5 87.886  \n\u0026#8195;\u0026#8195;Acc@1 67.320 Acc@5 87.812  \n\u0026#8195;\u0026#8195;Acc@1 67.404 Acc@5 87.866  \n\u0026#8195;finetune:  \n\u0026#8195;\u0026#8195;Acc@1 77.252 Acc@5 93.598  \n\u0026#8195;\u0026#8195;Acc@1 76.902 Acc@5 93.478  \n\u0026#8195;\u0026#8195;Acc@1 77.028 Acc@5 93.550  \n\u0026#8195;\u0026#8195;Acc@1 77.114 Acc@5 93.582  \n\u0026#8195;coco:  \n\u0026#8195;\u0026#8195;bbox: 38.9088,58.6155,42.1195,22.5249,43.4853,53.2623  \n\u0026#8195;\u0026#8195;segm: 34.1413,55.4126,36.3194,15.2867,37.2440,51.9860  \n\u0026#8195;\u0026#8195;bbox: 38.1508,57.6392,41.1611,20.7222,42.8992,51.8489  \n\u0026#8195;\u0026#8195;segm: 33.4272,54.5981,35.3582,14.3102,36.7986,50.7192  \n\u0026#8195;\u0026#8195;bbox: 38.7340,58.1209,42.1626,22.4818,43.5662,52.4255  \n\u0026#8195;\u0026#8195;segm: 33.9001,54.8911,36.1609,15.3182,37.3886,50.5798  \n\u0026#8195;\u0026#8195;bbox: 38.9785,58.5592,42.1268,22.3429,43.7718,52.9737  \n\u0026#8195;\u0026#8195;segm: 34.0710,55.2887,36.2458,15.4479,37.4157,50.8065  \n\u0026#8195;deeplab:  \n\u0026#8195;\u0026#8195;78.2688,58.6464,90.2149,77.6179  \n\u0026#8195;\u0026#8195;77.7682,57.8095,90.2813,77.4702  \n\u0026#8195;\u0026#8195;78.1918,58.4643,90.2833,77.6382  \n\u0026#8195;\u0026#8195;77.3166,58.1030,90.2396,77.6255  \n   \n+fast-moco:   \n\u0026#8195;linear:  \n\u0026#8195;\u0026#8195;Acc@1 70.778 Acc@5 89.818  \n\u0026#8195;\u0026#8195;Acc@1 70.858 Acc@5 89.918  \n\u0026#8195;\u0026#8195;Acc@1 70.868 Acc@5 89.872  \n\u0026#8195;\u0026#8195;Acc@1 70.854 Acc@5 89.946  \n\u0026#8195;finetune:  \n\u0026#8195;\u0026#8195;Acc@1 77.244 Acc@5 93.468  \n\u0026#8195;\u0026#8195;Acc@1 77.214 Acc@5 93.490  \n\u0026#8195;\u0026#8195;Acc@1 77.122 Acc@5 93.524  \n\u0026#8195;\u0026#8195;Acc@1 77.096 Acc@5 93.560  \n\u0026#8195;coco:  \n\u0026#8195;\u0026#8195;bbox: 38.8650,58.7779,41.7332,22.4263,43.9251,51.6782  \n\u0026#8195;\u0026#8195;segm: 33.9814,55.3545,35.9995,15.5256,37.7407,50.4110  \n\u0026#8195;\u0026#8195;bbox: 39.6814,59.4448,42.9916,22.3039,44.6076,53.6434  \n\u0026#8195;\u0026#8195;segm: 34.6912,56.1206,36.8248,15.7709,38.3276,51.9963  \n\u0026#8195;\u0026#8195;bbox: 39.4916,59.3736,42.8254,23.1930,44.4368,52.5870  \n\u0026#8195;\u0026#8195;segm: 34.5428,56.0413,36.6260,15.7507,38.4682,51.1193  \n\u0026#8195;\u0026#8195;bbox: 39.1963,59.0715,42.3853,22.1433,44.3828,52.2242  \n\u0026#8195;\u0026#8195;segm: 34.3486,55.7250,36.5795,15.4824,38.0844,50.9105  \n\u0026#8195;deeplab:  \n\u0026#8195;\u0026#8195;78.1100,59.0041,90.3268,78.1409   \n\u0026#8195;\u0026#8195;78.2087,59.1489,90.3703,78.0988  \n\u0026#8195;\u0026#8195;77.5934,58.0122,90.3006,77.9366  \n\u0026#8195;\u0026#8195;77.8860,58.6174,90.3806,78.2742  \n    \n\n+cutmix:   \n\u0026#8195;linear:  \n\u0026#8195;\u0026#8195;Acc@1 71.328 Acc@5 90.156  \n\u0026#8195;\u0026#8195;Acc@1 71.304 Acc@5 90.140   \n\u0026#8195;\u0026#8195;Acc@1 71.420 Acc@5 90.122  \n\u0026#8195;\u0026#8195;Acc@1 71.244 Acc@5 90.138  \n\u0026#8195;finetune:  \n\u0026#8195;\u0026#8195;Acc@1 77.144 Acc@5 93.610   \n\u0026#8195;\u0026#8195;Acc@1 77.012 Acc@5 93.440  \n\u0026#8195;\u0026#8195;Acc@1 77.284 Acc@5 93.570  \n\u0026#8195;\u0026#8195;Acc@1 77.208 Acc@5 93.564  \n\u0026#8195;coco:  \n\u0026#8195;\u0026#8195;bbox: 39.1084,59.1479,42.2791,22.4279,44.4237,53.2122  \n\u0026#8195;\u0026#8195;segm: 34.2483,55.8341,36.5099,15.1468,38.1949,51.5365  \n\u0026#8195;\u0026#8195;bbox: 39.5533,59.3614,42.9080,22.7960,44.7712,53.9648  \n\u0026#8195;\u0026#8195;segm: 34.5953,55.9939,36.8683,15.2826,38.3724,52.1124  \n\u0026#8195;\u0026#8195;bbox: 39.4069,59.3092,42.6750,23.1086,44.6547,53.2479  \n\u0026#8195;\u0026#8195;segm: 34.5314,55.9698,36.7320,16.6169,38.6138,51.2430  \n\u0026#8195;\u0026#8195;bbox: 39.5974,59.5217,42.5368,23.4857,45.2232,53.7143  \n\u0026#8195;\u0026#8195;segm: 34.5555,56.2276,36.6526,16.6227,38.6538,52.0167  \n\u0026#8195;deeplab:  \n\u0026#8195;\u0026#8195;78.6545,58.8794,90.4519,78.2854  \n\u0026#8195;\u0026#8195;78.5601,59.4348,90.4387,78.0768  \n\u0026#8195;\u0026#8195;78.3252,59.1720,90.4460,78.2306  \n\u0026#8195;\u0026#8195;78.9993,59.0939,90.5852,78.4380  \n\n\n+mixup:  \n\u0026#8195;linear:  \n\u0026#8195;\u0026#8195;Acc@1 70.426 Acc@5 89.920  \n\u0026#8195;\u0026#8195;Acc@1 70.502 Acc@5 89.952  \n\u0026#8195;\u0026#8195;Acc@1 70.458 Acc@5 89.982  \n\u0026#8195;\u0026#8195;Acc@1 70.292 Acc@5 89.952  \n\u0026#8195;finetune:  \n\u0026#8195;\u0026#8195;Acc@1 77.232 Acc@5 93.526  \n\u0026#8195;\u0026#8195;Acc@1 77.362 Acc@5 93.634  \n\u0026#8195;\u0026#8195;Acc@1 77.262 Acc@5 93.532  \n\u0026#8195;\u0026#8195;Acc@1 77.280 Acc@5 93.696  \n\u0026#8195;coco:  \n\u0026#8195;\u0026#8195;bbox: 39.4056,59.0497,42.3511,22.5197,44.3473,53.2681  \n\u0026#8195;\u0026#8195;segm: 34.5090,55.8374,36.8892,15.4050,38.3349,51.3551  \n\u0026#8195;\u0026#8195;bbox: 39.4914,59.2288,42.7277,21.8810,44.7128,53.6556  \n\u0026#8195;\u0026#8195;segm: 34.6709,56.1226,37.1046,15.4873,38.2889,52.5108  \n\u0026#8195;\u0026#8195;bbox: 39.4731,59.2949,42.4782,23.3873,44.6141,53.2002  \n\u0026#8195;\u0026#8195;segm: 34.6118,56.0852,36.8488,16.4710,38.2405,51.8747  \n\u0026#8195;\u0026#8195;bbox: 39.3198,58.9063,42.3989,22.9052,44.1219,53.4169  \n\u0026#8195;\u0026#8195;segm: 34.4959,55.8026,36.7658,16.2410,38.2230,52.2665  \n\u0026#8195;deeplab:  \n\u0026#8195;\u0026#8195;78.7261,58.7874,90.4714,78.3816  \n\u0026#8195;\u0026#8195;78.6566,58.2874,90.5395,78.3319  \n\u0026#8195;\u0026#8195;78.3647,58.4627,90.4798,78.5871  \n\u0026#8195;\u0026#8195;78.4664,58.7298,90.4983,78.3792  \n\n\n+dense:   \n\u0026#8195;linear:  \n\u0026#8195;\u0026#8195;Acc@1 68.878 Acc@5 88.988  \n\u0026#8195;\u0026#8195;Acc@1 68.794 Acc@5 88.962  \n\u0026#8195;\u0026#8195;Acc@1 68.722 Acc@5 88.982  \n\u0026#8195;\u0026#8195;Acc@1 68.784 Acc@5 88.952  \n\u0026#8195;finetune:  \n\u0026#8195;\u0026#8195;Acc@1 77.108 Acc@5 93.560  \n\u0026#8195;\u0026#8195;Acc@1 77.374 Acc@5 93.696  \n\u0026#8195;\u0026#8195;Acc@1 77.274 Acc@5 93.560  \n\u0026#8195;\u0026#8195;Acc@1 77.404 Acc@5 93.630  \n\u0026#8195;coco:  \n\u0026#8195;\u0026#8195;bbox: 39.9832,59.9284,43.4874,22.5677,45.1365,54.0637  \n\u0026#8195;\u0026#8195;segm: 34.9129,56.6404,37.0795,15.4479,38.6572,52.1838  \n\u0026#8195;\u0026#8195;bbox: 39.7174,59.5656,42.8223,22.8941,45.0750,53.2531  \n\u0026#8195;\u0026#8195;segm: 34.5298,56.3506,36.8583,15.6896,38.3153,51.9833  \n\u0026#8195;\u0026#8195;bbox: 40.1698,59.9970,43.3704,23.9472,45.6143,53.9589  \n\u0026#8195;\u0026#8195;segm: 34.9143,56.3698,37.4172,17.2151,38.8261,52.0241  \n\u0026#8195;\u0026#8195;bbox: 40.1558,59.7698,43.3668,22.2897,45.6031,54.1690  \n\u0026#8195;\u0026#8195;segm: 34.9127,56.6232,37.0422,15.5800,38.7052,52.5246  \n\u0026#8195;deeplab:  \n\u0026#8195;\u0026#8195;78.4191,58.8781,90.5701,78.7800  \n\u0026#8195;\u0026#8195;78.7982,59.7372,90.4969,78.4392  \n\u0026#8195;\u0026#8195;78.9305,59.1239,90.5806,78.7365  \n\u0026#8195;\u0026#8195;78.6499,58.9398,90.4527,78.1707  \n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcoincheung%2Fselfsup","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fcoincheung%2Fselfsup","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcoincheung%2Fselfsup/lists"}