{"id":13577250,"url":"https://github.com/cjf8899/KeepAugment_Pytorch","last_synced_at":"2025-04-05T11:31:48.713Z","repository":{"id":108884328,"uuid":"499000188","full_name":"cjf8899/KeepAugment_Pytorch","owner":"cjf8899","description":":low_brightness:Unofficial PyTorch implementation of  KeepAugment","archived":false,"fork":false,"pushed_at":"2022-06-06T07:30:07.000Z","size":61,"stargazers_count":22,"open_issues_count":1,"forks_count":1,"subscribers_count":1,"default_branch":"main","last_synced_at":"2024-11-05T14:43:56.090Z","etag":null,"topics":["augmentation","implementation","keepaugment","pytorch","saliency-map"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/cjf8899.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"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":"2022-06-02T05:14:45.000Z","updated_at":"2024-09-06T01:41:41.000Z","dependencies_parsed_at":"2023-03-13T14:21:56.920Z","dependency_job_id":null,"html_url":"https://github.com/cjf8899/KeepAugment_Pytorch","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/cjf8899%2FKeepAugment_Pytorch","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/cjf8899%2FKeepAugment_Pytorch/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/cjf8899%2FKeepAugment_Pytorch/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/cjf8899%2FKeepAugment_Pytorch/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/cjf8899","download_url":"https://codeload.github.com/cjf8899/KeepAugment_Pytorch/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":247331105,"owners_count":20921733,"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":["augmentation","implementation","keepaugment","pytorch","saliency-map"],"created_at":"2024-08-01T15:01:19.773Z","updated_at":"2025-04-05T11:31:48.476Z","avatar_url":"https://github.com/cjf8899.png","language":"Python","readme":"# KeepAugment_Pytorch\n\n\n## Unofficial implementation of [\"KeepAugment: A Simple Information-Preserving Data Augmentation Approach\"](https://openaccess.thecvf.com/content/CVPR2021/papers/Gong_KeepAugment_A_Simple_Information-Preserving_Data_Augmentation_Approach_CVPR_2021_paper.pdf). CVPR2021\n\n\u003cp align=\"center\"\u003e\u003cimg src=\"https://user-images.githubusercontent.com/53032349/171586571-6de784a5-0e34-4d63-9034-0fab7e62f69e.png\" width=\"100%\" height=\"100%\" title=\"70px\" alt=\"memoryblock\"\u003e\u003c/p\u003e\u003cbr\u003e\n\n## Results \n\n### CIFAR 10\n\n|              Model            |     ResNet-18     |  Wide ResNet-28-10  |\n| :---------------------------: | :---------------: | :-----------------: |\n| Cutout                        |  95.6±0.1(paepr)  |  96.9±0.1(paepr)    |\n| KeepCutout                    |        96.2       |        97.2         |\n| KeepCutout (low resolution)   |        96.1       |        97.1         |\n| KeepCutout (early loss)       |        96.2       |        TODO         |\n| KeepCutout (low + early)      |        96.2       |        TODO         |\n\n|              Model              |     ResNet-18     |  Wide ResNet-28-10  |\n| :-----------------------------: | :---------------: | :-----------------: |\n| AutoAugment                     |       96.0        |  97.3±0.1(paepr)    |\n| KeepAutoAugment                 |       96.4        |          97.4       |\n| KeepAutoAugment (low resolution)|       96.3        |          97.5       |\n| KeepAutoAugment (early loss)    |       96.3        |          97.4       |\n| KeepAutoAugment (low + early)   |       96.5        |         TODO        |\n\nAll results have slight differences(±0.1).\n\n## Run\n\nThe type of method is \u003cbr\u003e\n'keep_cutout', 'keep_cutout_low', 'keep_cutout_early', 'keep_cutout_low_early',\u003cbr\u003e\n'keep_autoaugment', 'keep_autoaugment_low', 'keep_autoaugment_early', 'keep_autoaugment_low_early'. \n\nThe type of model is 'resnet', 'wide_resnet'.\n\n```Shell\npython train.py --model resnet --method keep_cutout\n```\n\n### Any feedback on code simplification and incorrect implementation would be appreciated!\n\n## Referenced. Thank you all:+1:\nbaseline \u0026 cutout code : https://github.com/uoguelph-mlrg/Cutout\u003cbr\u003e\nautoaugment code : https://github.com/DeepVoltaire/AutoAugment\u003cbr\u003e\nsaliency map code : https://github.com/sunnynevarekar/pytorch-saliency-maps\u003cbr\u003e\n","funding_links":[],"categories":["Python"],"sub_categories":[],"project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcjf8899%2FKeepAugment_Pytorch","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fcjf8899%2FKeepAugment_Pytorch","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcjf8899%2FKeepAugment_Pytorch/lists"}