{"id":20305527,"url":"https://github.com/vitae-transformer/remote-sensing-rvsa","last_synced_at":"2025-04-05T17:09:27.274Z","repository":{"id":58976916,"uuid":"520325311","full_name":"ViTAE-Transformer/Remote-Sensing-RVSA","owner":"ViTAE-Transformer","description":"The official repo for [TGRS'22] \"Advancing Plain Vision Transformer Towards Remote Sensing Foundation Model\"","archived":false,"fork":false,"pushed_at":"2023-12-20T21:15:11.000Z","size":4716,"stargazers_count":400,"open_issues_count":24,"forks_count":30,"subscribers_count":6,"default_branch":"main","last_synced_at":"2024-08-07T13:28:16.654Z","etag":null,"topics":["deep-learning","foundation-model","foundation-models","object-detection","pytorch","remote-sensing","remote-sensing-foundation-model","scene-classification","self-supervised-learning","semantic-segmentation","transfer-learning","vision-transformer"],"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/ViTAE-Transformer.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":"2022-08-02T02:28:50.000Z","updated_at":"2024-08-05T05:11:46.000Z","dependencies_parsed_at":"2023-02-16T04:00:39.264Z","dependency_job_id":"785542cc-06a5-4590-853b-53b2aa0f4d53","html_url":"https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA","commit_stats":{"total_commits":99,"total_committers":1,"mean_commits":99.0,"dds":0.0,"last_synced_commit":"36c4c28c40f167c1bff8183a1c5642513765c4e2"},"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ViTAE-Transformer%2FRemote-Sensing-RVSA","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ViTAE-Transformer%2FRemote-Sensing-RVSA/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ViTAE-Transformer%2FRemote-Sensing-RVSA/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ViTAE-Transformer%2FRemote-Sensing-RVSA/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/ViTAE-Transformer","download_url":"https://codeload.github.com/ViTAE-Transformer/Remote-Sensing-RVSA/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":247369953,"owners_count":20927928,"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":["deep-learning","foundation-model","foundation-models","object-detection","pytorch","remote-sensing","remote-sensing-foundation-model","scene-classification","self-supervised-learning","semantic-segmentation","transfer-learning","vision-transformer"],"created_at":"2024-11-14T17:08:49.783Z","updated_at":"2025-04-05T17:09:27.255Z","avatar_url":"https://github.com/ViTAE-Transformer.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Advancing Plain Vision Transformer Towards Remote Sensing Foundation Model \n### Di Wang, Qiming Zhang, Yufei Xu, Jing Zhang, Bo Du, Dacheng Tao and Liangpei Zhang\n\n\u003cp align=\"center\"\u003e\n  \u003ca href=\"#updates\"\u003eUpdates\u003c/a\u003e |\n  \u003ca href=\"#introduction\"\u003eIntroduction\u003c/a\u003e |\n  \u003ca href=\"#results-and-models\"\u003eResults \u0026 Models\u003c/a\u003e |\n  \u003ca href=\"#usage\"\u003eUsage\u003c/a\u003e |\n\u003c/p \u003e\n\n\u003cp align=\"left\"\u003e\n\u003ca href=\"https://arxiv.org/abs/2208.03987\"\u003e\u003cimg src=\"https://img.shields.io/badge/arXiv-Paper-\u003ccolor\u003e\"\u003e\u003c/a\u003e\n\u003ca href=\"https://ieeexplore.ieee.org/document/9956816\"\u003e\u003cimg src=\"https://img.shields.io/badge/TGRS-Paper-blue\"\u003e\u003c/a\u003e\n\u003c/p\u003e\n\n[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/advancing-plain-vision-transformer-towards/object-detection-in-aerial-images-on-dota-1)](https://paperswithcode.com/sota/object-detection-in-aerial-images-on-dota-1?p=advancing-plain-vision-transformer-towards)\n[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/advancing-plain-vision-transformer-towards/object-detection-in-aerial-images-on-dior-r)](https://paperswithcode.com/sota/object-detection-in-aerial-images-on-dior-r?p=advancing-plain-vision-transformer-towards)\n[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/advancing-plain-vision-transformer-towards/aerial-scene-classification-on-ucm-50-as)](https://paperswithcode.com/sota/aerial-scene-classification-on-ucm-50-as?p=advancing-plain-vision-transformer-towards)\n[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/advancing-plain-vision-transformer-towards/aerial-scene-classification-on-aid-20-as)](https://paperswithcode.com/sota/aerial-scene-classification-on-aid-20-as?p=advancing-plain-vision-transformer-towards)\n[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/advancing-plain-vision-transformer-towards/aerial-scene-classification-on-aid-50-as)](https://paperswithcode.com/sota/aerial-scene-classification-on-aid-50-as?p=advancing-plain-vision-transformer-towards)\n[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/advancing-plain-vision-transformer-towards/aerial-scene-classification-on-nwpu-10-as)](https://paperswithcode.com/sota/aerial-scene-classification-on-nwpu-10-as?p=advancing-plain-vision-transformer-towards)\n[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/advancing-plain-vision-transformer-towards/aerial-scene-classification-on-nwpu-20-as)](https://paperswithcode.com/sota/aerial-scene-classification-on-nwpu-20-as?p=advancing-plain-vision-transformer-towards)\n[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/advancing-plain-vision-transformer-towards/semantic-segmentation-on-isprs-potsdam)](https://paperswithcode.com/sota/semantic-segmentation-on-isprs-potsdam?p=advancing-plain-vision-transformer-towards)\n[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/advancing-plain-vision-transformer-towards/semantic-segmentation-on-isaid)](https://paperswithcode.com/sota/semantic-segmentation-on-isaid?p=advancing-plain-vision-transformer-towards)\n[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/advancing-plain-vision-transformer-towards/semantic-segmentation-on-loveda)](https://paperswithcode.com/sota/semantic-segmentation-on-loveda?p=advancing-plain-vision-transformer-towards)\n\n## Current applications\n\n\u003e **ViTAE: Please see [ViTAE-Transformer](https://github.com/ViTAE-Transformer/ViTAE-Transformer)**;\n\n\u003e **VSA: Please see [ViTAE-VSA](https://github.com/ViTAE-Transformer/ViTAE-VSA)**;\n\n\u003e **Matting: Please see [ViTAE-Transformer for matting](https://github.com/ViTAE-Transformer/ViTAE-Transformer-Matting)**;\n\n\u003e **Remote Sensing Pretraining: Please see [ViTAE-Transformer-Remote-Sensing](https://github.com/ViTAE-Transformer/ViTAE-Transformer-Remote-Sensing)**;\n\n## Updates\n\n### 2025.02.24\n\nThe classification inference framework of RVSA on the Huawei MindSpore is provided! See **/MAEPretrain_SceneClassification/MindSpore/**\n\n### 2025.02.24\n\nThe classification inference framework of RVSA on the Tsinghua Jittor is provided! See **/MAEPretrain_SceneClassification/jittor/**\n\n### 2025.02.16\n\nThe classification inference framework of RVSA on the Baidu PaddlePaddle is provided! See **/MAEPretrain_SceneClassification/Paddle/**\n\n### 2023.10.18\n\nRVSA won the highly cited paper!\n\n\u003cfigure\u003e\n\u003cdiv align=\"center\"\u003e\n\u003cimg src=highlycited.png width=\"100%\"\u003e\n\u003c/div\u003e\n\u003c/figure\u003e\n\n### 2023.03.18\nViTAE-B + RVSA helped us win the championship of \"High Resolution SAR Image Coastal Aquaculture Farm Segmentation Track\" in \"The 5th Gaofen Challenge\", Team: TNT. （第五届“中科星图杯”国际高分遥感图像解译大赛高分辨率SAR图像中近海养殖场分割赛道冠军）[News](https://mp.weixin.qq.com/s/zJ2jOXa8rwIPGJ0vg2znDw)\n\n### 2023.01.18\n\nOur models have been supported by [LuoJiaNET](https://github.com/WHULuoJiaTeam/luojianet), please refer to [RS-Vision-Foundation-Models](https://github.com/WHULuoJiaTeam/Model_Zoo/tree/main/RS_Vision_Foundation_Models) for more details.\n\n### 2022.11.21\n\nThe early access is available! [TGRS link](https://ieeexplore.ieee.org/document/9956816)\n\n### 2022.11.15\n\nThe arXiv has been updated! [arXiv link](https://arxiv.org/abs/2208.03987)\n\n### 2022.11.06\n\nThe paper has been accepted by IEEE TGRS!\n\n### 2022.10.11 \n\nThe codes, configs and training logs of segmentation in fintuning are released!\n\n### 2022.10.09 \n\nThe codes, configs and training logs of detection in fintuning are released!\n\n### 2022.10.08 \n\nThe codes of pretraining and classification in fintuning are released!\n\n### 2022.09.19 \n\nThe codes and training logs of the [VSA](https://github.com/ViTAE-Transformer/ViTAE-VSA) have been released, which is the foundation of our RVSA.\n\n## Introduction\n\nThis repository contains codes, models and test results for the paper \"[Advancing Plain Vision Transformer Towards Remote Sensing Foundation Model](https://arxiv.org/abs/2208.03987)\".\n\nWe resort to plain vision transformers with about 100M and make the first attempt to propose large vision models customized for RS tasks and propose a new rotated varied-size window attention (RVSA) to substitute the original full attention to handle the large image size and objects of various orientations in RS images. The RVSA could significantly reduce the computational cost and memory footprint while learn better object representation by extracting rich context from the generated diverse windows.\n\n\u003cfigure\u003e\n\u003cimg src=Figs/framework.png\u003e\n\u003cfigcaption align = \"center\"\u003e\u003cb\u003eFig.1 - The pipeline of pretraining and finetuning. \u003c/b\u003e\u003c/figcaption\u003e\n\u003c/figure\u003e\n\n\u0026emsp;\n\n\u003cfigure\u003e\n\u003cimg src=Figs/vit_rvsa.png\u003e\n\u003cfigcaption align = \"center\"\u003e\u003cb\u003eFig.2 - The structure and block of the adopted plain vision transformer, and the proposed RVSA. \u003c/b\u003e\u003c/figcaption\u003e\n\u003c/figure\u003e\n\n\n## Results and Models\n\n### Pretraining \n\n#### MillionAID\n|Pretrain|Backbone | Input size | Params (M) | Pretrained model|\n|-------|-------- | ----------  | ----- | ----- |\n| MAE | ViT-B | 224 × 224 | 86| [Weights](https://1drv.ms/u/s!AimBgYV7JjTlgUPBC6cvpo4oZDSR?e=kNCAhO) |\n| MAE | ViTAE-B | 224 × 224 | 89 | [Weights](https://1drv.ms/u/s!AimBgYV7JjTlgUIde2jzcjrrWasP?e=gyLn29) |\n\n### Object Detection\n\n#### DOTA-V1.0 Single-Scale\n| Method | Pretrain | Backbone | Lr schd | mAP | Config | Log | Model |\n| ------ |----------| -------- | --------- | ------- | :---: | :------: | :---: |\n| Oriented R-CNN | MAE | ViT-B + RVSA | 1x | 78.75 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_wsz7_fpn_1x_dota10_lr1e-4_ldr75_dpr15.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/logs/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_wsz7_fpn_1x_dota10_lr1e-4_ldr75_dpr15.log) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgVJM4Znng50US8KD?e=o4MRMQ) |\n| Oriented R-CNN | MAE | ViT-B + RVSA $^ \\Diamond$ | 1x | 78.61 |[Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dota10_lr1e-4_ldr75_dpr15.py)| [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/logs/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dota10_lr1e-4_ldr75_dpr15.log) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgVOYOQ3d_QS1Vkco?e=aOrzfJ) |  |\n| Oriented R-CNN | MAE | ViTAE-B + RVSA | 1x | 78.96  | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_wsz7_fpn_1x_dota10_lr1e-4_ldr75_dpr15.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/logs/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_wsz7_fpn_1x_dota10_lr1e-4_ldr75_dpr15.log) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgVAG59CJWkxA7f5U?e=hjR3Bx) |\n| Oriented R-CNN | MAE | ViTAE-B + RVSA $^ \\Diamond$ | 1x | 78.99 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dota10_lr1e-4_ldr75_dpr15.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/logs/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dota10_lr1e-4_ldr75_dpr15.log) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgVEPLn_D1ph3EQ5Q?e=MXLorq) |\n\n\n#### DOTA-V1.0 Multi-Scale\n| Method | Pretrain | Backbone | Lr schd | mAP | Config | Log | Model |\n| ------ |----------| -------- | --------- | ------- | :---: | :------: | :---: |\n| Oriented R-CNN | MAE | ViT-B + RVSA | 1x | 81.01 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_wsz7_fpn_1x_dota10_ms_lr1e-4_ldr75_dpr15.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/logs/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_wsz7_fpn_1x_dota10_ms_lr1e-4_ldr75_dpr15.log)| [Model](https://1drv.ms/u/s!AimBgYV7JjTlgU6SB-_d-xk5Fh5N?e=KSkQqA) |\n| Oriented R-CNN | MAE | ViT-B + RVSA $^ \\Diamond$ | 1x | 80.80 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dota10_ms_lr1e-4_ldr75_dpr15.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/logs/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dota10_ms_lr1e-4_ldr75_dpr15.log) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgccA-3Vs3J5SZqB0lg?e=0Wrast) |\n| Oriented R-CNN | MAE | ViTAE-B + RVSA | 1x | 81.24 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_wsz7_fpn_1x_dota10_ms_lr1e-4_ldr75_dpr15.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/logs/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_wsz7_fpn_1x_dota10_ms_lr1e-4_ldr75_dpr15.log) |  [Model](https://1drv.ms/u/s!AimBgYV7JjTlgccBdC967NGcR_FcvA?e=6q2Vd2) |\n| Oriented R-CNN | MAE | ViTAE-B + RVSA $^ \\Diamond$ | 1x | 81.18 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dota10_ms_lr1e-4_ldr75_dpr15.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/logs/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dota10_ms_lr1e-4_ldr75_dpr15.log) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgccCSURJPWl1jPdeIA?e=HtYQgD) |\n\n#### DIOR-R\n| Method | Pretrain | Backbone | Lr schd | mAP | Config | Log | Model |\n| ------ |----------| -------- | --------- | ------- | :---: | :------: | :---: |\n| Oriented R-CNN | MAE | ViT-B + RVSA | 1x | 70.67 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_wsz7_fpn_1x_dior_lr1e-4_ldr75_dpr15.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/logs/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_wsz7_fpn_1x_dior_lr1e-4_ldr75_dpr15.log) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgUjzxMV17pmGV-md?e=b0z0Gn) |\n| Oriented R-CNN | MAE | ViT-B + RVSA $^ \\Diamond$ | 1x | 70.85 |  [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dior_lr1e-4_ldr75_dpr15.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/logs/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dior_lr1e-4_ldr75_dpr15.log) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgUv2gHKLWOwa0_AD?e=gpXnKG) |\n| Oriented R-CNN | MAE | ViTAE-B + RVSA | 1x | 70.95 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_wsz7_fpn_1x_dior_lr1e-4_ldr75_dpr10.py) |[Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/logs/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_wsz7_fpn_3x_dior_lr1e-4_ldr75_dpr10.log) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgUqfEy2J7BTmKKlK?e=nsoWwM) |\n| Oriented R-CNN | MAE | ViTAE-B + RVSA $^ \\Diamond$ | 1x | 71.05 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dior_lr1e-4_ldr75_dpr10.py)| [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Object%20Detection/logs/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_kvdiff_wsz7_fpn_3x_dior_lr1e-4_ldr75_dpr10.log) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgUkvsiV0YJkWWyuY?e=IMHGAR) |\n\n### Scene Classification\n\n|Pretrain | Backbone | UCM-55 | AID-28 | AID-55 | NWPU-19 | NWPU-28 |\n|----------|-------- | --------- | ------- | --- | ------ | --- | \n| MAE | ViT-B + RVSA | 99.70 | 96.92 | 98.33 | 93.79 | 95.49 |\n|     |              |[Model](https://1drv.ms/u/s!AimBgYV7JjTlgWwwNEk-zZN8Zddb?e=RIfcn9) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgXaE3XBegpd9Awqb?e=snF1gg) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgXvqgHU6iJ4aJ0L1?e=XXLbqy) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgW84yjf3C-RUNmOc?e=LYJLpB) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgXPYuJWRpRRzu9B_?e=NJjcXB) |\n| MAE | ViT-B + RVSA $^ \\Diamond$ | 99.58 | 96.86 | 98.44 | 93.74 | 95.45 |\n|     |              |[Model](https://1drv.ms/u/s!AimBgYV7JjTlgWnicUuVKBIGZAW0?e=xjfV8z) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgXhlvwoQP1Sbb9RG?e=gnMNTQ) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgXzoV-yx80lxcwR7?e=CL4k33) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgXBJyGduYfF5SCB7?e=LgdIXb) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgXROC5R_8d3CViay?e=UdJPVM) |\n| MAE | ViTAE-B + RVSA | 99.56 | 97.03 | 98.48 | 93.93 | 95.69 |\n|     |              |[Model](https://1drv.ms/u/s!AimBgYV7JjTlgWv-ct-hZwKbSqdg?e=tRgL4J) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgXWc6RWse1WxjuI2?e=UtGAm3) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgXkt_Lv6GljM4AkG?e=1wk8K3) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgW0vDOaTsch8Spqo?e=km6axJ) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgXLPtT4hVj4diQax?e=7c9k0U) |\n| MAE | ViTAE-B + RVSA $^ \\Diamond$ | 99.50 | 97.01 | 98.50 | 93.92 | 95.66|\n|     |              |[Model](https://1drv.ms/u/s!AimBgYV7JjTlgWqe8N6wagHiPiFe?e=7gIVAn) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgXch-D4GJ1Gutg3J?e=U3W9AJ) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgXocg8ael_cdmdhK?e=pPJGRm) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgW7bR8HfcpK4wig4?e=ZYOmdS) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgXFEJ_T2wZRVhiow?e=CQ6son) |\n\n### Semantic Segmentation\n\n#### ISPRS Potsdam\n\n| Method | Pretrain | Backbone | Crop size | Lr schd | OA | Config | Log | Model |\n| ------ | ----------|-------- | --------- | ------- | --- | :------: | :---: | :-----: |\n| UperNet| MAE | ViT-B + RVSA | 512 × 512 | 160k | 90.60 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/configs/vit_base_win/upernet_vit_base_win_rvsa_v3_512x512_160k_potsdam_rgb_dpr10_lr6e5_lrd90_ps16_class5_ignore5.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/logs/upernet_vit_base_win_rvsa_v3_512x512_160k_potsdam_rgb_dpr10_lr6e5_lrd90_ps16_class5_ignore5.log.json) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlggsyQuSY6EYcOj2s?e=0k1MBH) |\n| UperNet| MAE | ViT-B + RVSA $^ \\Diamond$ | 512 × 512 | 160k | 90.77 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/configs/vit_base_win/upernet_vit_base_win_rvsa_v3_kvdiff_512x512_160k_potsdam_rgb_dpr10_lr6e5_lrd90_ps16_class5_ignore5.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/logs/upernet_vit_base_win_rvsa_v3_kvdiff_512x512_160k_potsdam_rgb_dpr10_lr6e5_lrd90_ps16_class5_ignore5.log.json) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlggyehoVuz4A7i9A6?e=WehMl8) |\n| UperNet| MAE | ViTAE-B + RVSA | 512 × 512 | 160k | 91.22 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/configs/vit_base_win/upernet_vitae_nc_base_rvsa_v3_wsz7_512x512_160k_potsdam_rgb_dpr10_lr6e5_lrd90_ps16_class5_ignore5.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/logs/upernet_vitae_nc_base_rvsa_v3_wsz7_512x512_160k_potsdam_rgb_dpr10_lr6e5_lrd90_ps16_class5_ignore5.log.json) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlggmAnAEe7hb5p8id?e=16oojH) |\n| UperNet| MAE | ViTAE-B + RVSA $^ \\Diamond$ | 512 × 512 | 160k | 91.15 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/configs/vit_base_win/upernet_vitae_nc_base_rvsa_v3_kvdiff_wsz7_512x512_160k_potsdam_rgb_dpr10_lr6e5_lrd90_ps16_class5_ignore5.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/logs/upernet_vitae_nc_base_rvsa_v3_kvdiff_wsz7_512x512_160k_potsdam_rgb_dpr10_lr6e5_lrd90_ps16_class5_ignore5.log.json) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlggqk2eQmyCF2E2Y2?e=hYxZQv) |\n\n#### iSAID\n\n| Method | Pretrain | Backbone | Crop size | Lr schd | mIOU | Config | Log | Model |\n| ------ | ----------|-------- | --------- | ------- | --- | :------: | :---: | :-----: |\n| UperNet| MAE | ViT-B + RVSA | 896 × 896 | 160k | 63.76 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/configs/vit_base_win/upernet_vit_base_win_rvsa_v3_wsz7_896x896_160k_isaid_dpr10_lr6e5_lrd90_ps16.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/logs/upernet_vit_base_win_rvsa_v3_wsz7_896x896_160k_isaid_dpr10_lr6e5_lrd90_ps16.log.json) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlggHtXbZ3T7hC-oqL?e=AGrYbz) |\n| UperNet| MAE | ViT-B + RVSA $^ \\Diamond$ | 896 × 896 | 160k | 63.85 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/configs/vit_base_win/upernet_vit_base_win_rvsa_v3_kvdiff_wsz7_896x896_160k_isaid_dpr10_lr6e5_lrd90_ps16.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/logs/upernet_vit_base_win_rvsa_v3_kvdiff_wsz7_896x896_160k_isaid_dpr10_lr6e5_lrd90_ps16.log.json) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlggQOfObKs86ZD-fd?e=Wz3MUe) |\n| UperNet| MAE | ViTAE-B + RVSA | 896 × 896 | 160k | 63.48 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/configs/vit_base_win/upernet_vitae_nc_base_rvsa_v3_wsz7_896x896_160k_isaid_dpr10_lr6e5_lrd90_ps16.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/logs/upernet_vitae_nc_base_rvsa_v3_wsz7_896x896_160k_isaid_dpr10_lr6e5_lrd90_ps16.log.json) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlggLl2cfm1Ro-ekra?e=EFCghU) |\n| UperNet| MAE | ViTAE-B + RVSA $^ \\Diamond$ | 896 × 896 | 160k | 64.49 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/configs/vit_base_win/upernet_vitae_nc_base_rvsa_v3_kvdiff_wsz7_896x896_160k_isaid_dpr10_lr6e5_lrd90_ps16.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/logs/upernet_vitae_nc_base_rvsa_v3_kvdiff_wsz7_896x896_160k_isaid_dpr10_lr6e5_lrd90_ps16.log.json) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlggPWSalF-CBKETHD?e=BRe1HQ) |\n\n#### LoveDA\n\n| Method | Pretrain | Backbone | Crop size | Lr schd | mIOU | Config | Log | Model |\n| ------ | ----------|-------- | --------- | ------- | --- | :------: | :---: | :-----: |\n| UperNet| MAE | ViT-B + RVSA | 512 × 512 | 160k | 51.95 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/configs/vit_base_win/upernet_vit_base_win_rvsa_v3_wsz7_512x512_160k_loveda_dpr10_lr6e5_lrd90_ps16.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/logs/upernet_vit_base_win_rvsa_v3_wsz7_512x512_160k_loveda_dpr10_lr6e5_lrd90_ps16.log.json) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlggcH-UQUNSDg8AMh?e=h1fiW0) |\n| UperNet| MAE | ViT-B + RVSA $^ \\Diamond$ | 512 × 512 | 160k | 51.95 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/configs/vit_base_win/upernet_vit_base_win_rvsa_v3_kvdiff_wsz7_512x512_160k_loveda_dpr10_lr6e5_lrd90_ps16.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/logs/upernet_vit_base_win_rvsa_v3_kvdiff_wsz7_512x512_160k_loveda_dpr10_lr6e5_lrd90_ps16.log.json) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlgghifTHhpfc5kPZW?e=HjV8Ib) |\n| UperNet| MAE | ViTAE-B + RVSA | 512 × 512 | 160k | 52.26 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/configs/vit_base_win/upernet_vitae_nc_base_rvsa_v3_wsz7_512x512_160k_loveda_dpr10_lr6e5_lrd90_ps16.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/logs/upernet_vitae_nc_base_rvsa_v3_wsz7_512x512_160k_loveda_dpr10_lr6e5_lrd90_ps16.log.json) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlggX66jVrAoWdZrK8?e=8hZsse) |\n| UperNet| MAE | ViTAE-B + RVSA $^ \\Diamond$ | 512 × 512 | 160k | 52.44 | [Config](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/configs/vit_base_win/upernet_vitae_nc_base_rvsa_v3_kvdiff_wsz7_512x512_160k_loveda_dpr10_lr6e5_lrd90_ps16.py) | [Log](https://github.com/ViTAE-Transformer/Remote-Sensing-RVSA/blob/main/Semantic%20Segmentation/logs/upernet_vitae_nc_base_rvsa_v3_kvdiff_wsz7_512x512_160k_loveda_dpr10_lr6e5_lrd90_ps16.log.json) | [Model](https://1drv.ms/u/s!AimBgYV7JjTlggaDnTijEqVrAHgg?e=CJHcTc) |\n\n## Usage\n\nEnvironment:\n\n- Python 3.8.5\n- Pytorch 1.9.0+cu111\n- torchvision 0.10.0+cu111\n- timm 0.4.12\n- mmcv-full 1.4.1\n\n### Pretraining \u0026 Finetuning-Classification\n\n#### Pretraining (8 × A100 GPUs, 3~5 days)\n\n1. Preparing the MillionAID: Download the [MillionAID](https://captain-whu.github.io/DiRS/). Here, we use previous `train_labels.txt` and `valid_labels.txt` of the [RSP](https://github.com/ViTAE-Transformer/RSP.git), which contain labels. However, since we conduct the ***unsupervised pretraining***, the labels are not necessary. It is easy for users to record image names and revise corresponding codes `MAEPretrain_SceneClassification/util/datasets.py/class MillionAIDDataset`.\n\n2. Pretraining: take ViT-B as an example (batchsize: 2048=8*256)\n\n```\npython -m torch.distributed.launch --nproc_per_node 8 --master_port 10000 main_pretrain.py \\\n--dataset 'millionAID' --model 'mae_vit_base_patch16' \\\n--batch_size 256 --epochs 1600 --warmup_epochs 40 \\\n--input_size 224 --mask_ratio 0.75 \\\n--blr 1.5e-4  --weight_decay 0.05 --gpu_num 8 \\\n--output_dir '../mae-main/output/'\n```\n*Note: Padding the convolutional kernel of PCM in the pretrained ViTAE-B with `convertK1toK3.py` for finetuning.*\n\n3. Linear probe: an example of evaluating the pretrained ViT-B on UCM-55\n\n```\nCUDA_VISIBLE_DEVICES=0 python -m torch.distributed.launch --nproc_per_node 1 --master_port 10000 main_linprobe.py \\\n--dataset 'ucm' --model 'vit_base_patch16' \\\n--batch_size 256 --epochs 100 --warmup_epochs 10 \\\n--blr 1e-1  --weight_decay 0 --tag 0 \\\n--finetune '../mae-main/output/millionAID_224/1600_0.75_0.00015_0.05_2048/checkpoint-1599.pth'\n```\n\n#### Finetuning evaluation for pretraining \u0026 Finetuning-Classification\n\nFor instance, finetuning ViTAE-B + RVSA on NWPU-28\n\n```\nCUDA_VISIBLE_DEVICES=0 python -m torch.distributed.launch --nproc_per_node 1 --master_port 20000 main_finetune.py \\\n--dataset 'nwpu' --model 'vitae_nc_base_win_rvsa' --input_size 224 --postfix 'sota' \\\n--batch_size 64 --epochs 200 --warmup_epochs 5 \\\n--blr 1e-3  --weight_decay 0.05 --split 28 --tag 0 --exp_num 1 \\\n--finetune '../mae-main/output/mae_vitae_base_pretrn/millionAID_224/1600_0.75_0.00015_0.05_2048/checkpoint-1599-transform-no-average.pth'\n```\n\n### Finetuning-Detection \u0026 Finetuning-Segmentation\n\nSince we use OBBDetection and MMSegmenation to implement corresponding detection or segmentation models, we only provide necessary config and backbone files. The main frameworks are both in [RSP](https://github.com/ViTAE-Transformer/RSP.git)\n\n```\ngit clone https://github.com/ViTAE-Transformer/ViTAE-Transformer-Remote-Sensing.git\n```\n\nThe installation and dataset preparation can separately refer [OBBDetection-installation](https://github.com/jbwang1997/OBBDetection/blob/master/docs/install.md) and \n[MMSegmentation-installation](https://github.com/open-mmlab/mmsegmentation/blob/master/docs/en/get_started.md#installation)\n\nThen put these files into corresponding folders.\n\nFor convenience, we preserve the relative path for users to find files.\n\nFor example, put `./Object Detection/mmdet/models/backbones/vit_win_rvsa_v3_wsz7.py` into `ViTAE-Transformer-Remote-Sensing/Object Detection/mmdet/models/backbones`\n\n#### Training-Detection\n\nFirst, `cd ./Object Detection` \n\nThen, we provide several examples. For instance, \n\nTraining the Oriented-RCNN with ViT-B + RVSA on DOTA-V1.0 multi-scale detection dataset with 2 GPUs\n\n```\nCUDA_VISIBLE_DEVICES=0,1 python -m torch.distributed.launch --nproc_per_node=2 --master_port=40000 tools/train.py \\\nconfigs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_wsz7_fpn_1x_dota10_ms_lr1e-4_ldr75_dpr15.py \\\n--launcher 'pytorch' --options 'find_unused_parameters'=True\n```\n\nTraining the Oriented-RCNN with ViTAE-B + RVSA $^ \\Diamond$ backbone on DIOR-R detection dataset with 1 GPU\n\n```\nCUDA_VISIBLE_DEVICES=0 python -m torch.distributed.launch --nproc_per_node=1 --master_port=40001 tools/train.py \\\nconfigs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dior_lr1e-4_ldr75_dpr10.py \\\n--launcher 'pytorch' --options 'find_unused_parameters'=True\n```\n\n#### Inference-Detection\n\nPredicting the saving detection map using ViT-B + RVSA $^ \\Diamond$ on DOTA-V1.0 scale-scale detection dataset\n\n```\nCUDA_VISIBLE_DEVICES=0 python tools/test.py configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dota10_lr1e-4_ldr75_dpr15.py \\\n../OBBDetection/work_dirs/faster/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dota10_lr1e-4_ldr75_dpr15/latest.pth \\\n--format-only --show-dir work_dirs/save/faster/display/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dota10_lr1e-4_ldr75_dpr15 \\\n--options save_dir='work_dirs/save/faster/full_det/faster_rcnn_orpn_our_rsp_vit-base-win-rvsa_v3_kvdiff_wsz7_fpn_1x_dota10_lr1e-4_ldr75_dpr15' nproc=1\n```\n\nEvaluating the detection maps predicted by ViTAE-B + RVSA on DIOR-R dataset\n\n```\nCUDA_VISIBLE_DEVICES=0 python tools/test.py configs/obb/oriented_rcnn/vit_base_win/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_wsz7_fpn_1x_dior_lr1e-4_ldr75_dpr10.py \\\n../OBBDetection/work_dirs/faster/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_wsz7_fpn_1x_dior_lr1e-4_ldr75_dpr10/latest.pth \\\n--out work_dirs/save/faster/full_det/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_wsz7_fpn_1x_dior_lr1e-4_ldr75_dpr10/det_result.pkl --eval 'mAP' \\\n--show-dir work_dirs/save/faster/display/faster_rcnn_orpn_our_rsp_vitae-nc-base-win-rvsa_v3_wsz7_fpn_1x_dior_lr1e-4_ldr75_dpr10\n```\n\n*Note: the pathes of saved maps and outputs should be constructed before evaluating the DIOR-R testing set.*\n\n#### Training \u0026 Evaluation-Segmentation\n\n`cd ./Semantic Segmentation` \n\nTraining and evaluation the UperNet with ViT-B + RVSA backbone on Potsdam dataset:\n\n```\nCUDA_VISIBLE_DEVICES=0,1 python -m torch.distributed.launch --nproc_per_node=2 --master_port=30000 tools/train.py \\\nconfigs/vit_base_win/upernet_vit_base_win_rvsa_v3_512x512_160k_potsdam_rgb_dpr10_lr6e5_lrd90_ps16_class5_ignore5.py \\\n--launcher 'pytorch' --cfg-options 'find_unused_parameters'=True\n```\n\n*Note: when training on the LoveDA, please add `--no-validate`*\n\nInference the LoveDA dataset for online evaluation using the UperNet with ViTAE-B + RVSA $^ \\Diamond$ backbone\n\n```\nCUDA_VISIBLE_DEVICES=0 python tools/test.py configs/vit_base_win/upernet_vitae_nc_base_rvsa_v3_kvdiff_wsz7_512x512_160k_loveda_dpr10_lr6e5_lrd90_ps16.py \\\n../mmsegmentation-master/work_dirs/upernet_vitae_nc_base_rvsa_v3_kvdiff_wsz7_512x512_160k_loveda_dpr10_lr6e5_lrd90_ps16/latest.pth \\\n--format-only --eval-options imgfile_prefix=\"work_dirs/display/upernet_vitae_nc_base_rvsa_v3_kvdiff_wsz7_512x512_160k_loveda_dpr10_lr6e5_lrd90_ps16/result\" \\\n--show-dir work_dirs/display/upernet_vitae_nc_base_rvsa_v3_kvdiff_wsz7_512x512_160k_loveda_dpr10_lr6e5_lrd90_ps16/rgb\n```\n\n### ***When finetuning with more than one GPU for detection or segmentation, please use `nn.SyncBatchNorm` in the NormalCell of ViTAE models.***\n\n## Citation\n\nIf this repo is useful for your research, please consider citation\n\n```\n@ARTICLE{rvsa,\n  author={Wang, Di and Zhang, Qiming and Xu, Yufei and Zhang, Jing and Du, Bo and Tao, Dacheng and Zhang, Liangpei},\n  journal={IEEE Transactions on Geoscience and Remote Sensing}, \n  title={Advancing Plain Vision Transformer Toward Remote Sensing Foundation Model}, \n  year={2023},\n  volume={61},\n  number={},\n  pages={1-15},\n  doi={10.1109/TGRS.2022.3222818}\n }\n\n@ARTICLE{rsp,\n  author={Wang, Di and Zhang, Jing and Du, Bo and Xia, Gui-Song and Tao, Dacheng},\n  journal={IEEE Transactions on Geoscience and Remote Sensing}, \n  title={An Empirical Study of Remote Sensing Pretraining}, \n  year={2023},\n  volume={61},\n  number={},\n  pages={1-20},\n  doi={10.1109/TGRS.2022.3176603}\n}\n```\n\n## Statement\n\nThis project is under MIT licence. For any other questions please contact [di.wang at gmail.com](mailto:wd74108520@gmail.com) .\n\n## References\n\nThe codes of Pretraining \u0026 Scene Classification part mainly from [MAE](https://github.com/facebookresearch/mae).\n\n## Relevant Projects\n[1] \u003cstrong\u003eAn Empirical Study of Remote Sensing Pretraining, IEEE TGRS, 2022\u003c/strong\u003e | [Paper](https://ieeexplore.ieee.org/document/9782149) | [Github](https://github.com/ViTAE-Transformer/ViTAE-Transformer-Remote-Sensing)\n\u003cbr\u003e\u003cem\u003e\u0026ensp; \u0026ensp; \u0026ensp;Di Wang\u003csup\u003e\u0026#8727;\u003c/sup\u003e, Jing Zhang\u003csup\u003e\u0026#8727;\u003c/sup\u003e, Bo Du, Gui-Song Xia and Dacheng Tao\u003c/em\u003e\n\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fvitae-transformer%2Fremote-sensing-rvsa","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fvitae-transformer%2Fremote-sensing-rvsa","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fvitae-transformer%2Fremote-sensing-rvsa/lists"}