{"id":13559579,"url":"https://github.com/jwwangchn/AI-TOD","last_synced_at":"2025-04-03T15:30:37.234Z","repository":{"id":37772792,"uuid":"303620675","full_name":"jwwangchn/AI-TOD","owner":"jwwangchn","description":"Official code for \"Tiny Object Detection in Aerial Images\".","archived":false,"fork":false,"pushed_at":"2024-11-13T20:00:00.000Z","size":2041,"stargazers_count":192,"open_issues_count":28,"forks_count":21,"subscribers_count":3,"default_branch":"master","last_synced_at":"2024-11-13T20:35:31.073Z","etag":null,"topics":["aerial-images","computer-vision","dataset","object-detection"],"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/jwwangchn.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}},"created_at":"2020-10-13T07:19:58.000Z","updated_at":"2024-11-13T20:00:03.000Z","dependencies_parsed_at":"2024-08-01T13:15:26.401Z","dependency_job_id":"9fe64bde-345b-4487-ac2c-2707dab98b02","html_url":"https://github.com/jwwangchn/AI-TOD","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/jwwangchn%2FAI-TOD","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jwwangchn%2FAI-TOD/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jwwangchn%2FAI-TOD/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jwwangchn%2FAI-TOD/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/jwwangchn","download_url":"https://codeload.github.com/jwwangchn/AI-TOD/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":247027672,"owners_count":20871571,"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":["aerial-images","computer-vision","dataset","object-detection"],"created_at":"2024-08-01T13:00:29.484Z","updated_at":"2025-04-03T15:30:37.207Z","avatar_url":"https://github.com/jwwangchn.png","language":"Python","funding_links":[],"categories":["Dataset","Optical Aerial Imagery Datasets","Anti-UAV Datasets"],"sub_categories":[],"readme":"# AI-TOD\n\n[[Paper]](https://drive.google.com/file/d/1IiTp7gilwDCGr8QR_H9Covz8aVK7LXiI/view?usp=sharing) AI-TOD is a dataset for tiny object detection in aerial images.\n\n\n[[Dataset]](https://github.com/jwwangchn/AI-TOD) Please download the [xView trainig set](http://xviewdataset.org/#dataset) and [AI-TOD_wo_xview](https://drive.google.com/drive/folders/1uNY_rcOO5LrWibXRY6l2dvqSbK6xikJp?usp=sharing) to [construct](aitodtoolkit) the complete AI-TOD dataset!\n\n![](demo/samples.png)\n\n## Description\n\nAI-TOD comes with 700,621 object instances for eight categories across 28,036 aerial images. Compared to existing object detection datasets in aerial images, the mean size of objects in AI-TOD is about 12.8 pixels, which is much smaller than others.\n\n\u003cimg src=\"demo/size_ratios.png\" width=\"500px\" div align=center /\u003e\n\n## Download \n\nYou need to download the following two parts (Part1: xView training set, Part2: part of AI-TOD) and use our end-to-end synthesis tool to generate the complete AI-TOD dataset. (Note the we have released the **complete annotations** of AI-TOD, you only need to **generate images**)\n* xView training set. [[Website]](http://xviewdataset.org/#dataset) \n* Part of AI-TOD. [[GoogleDrive]](https://drive.google.com/drive/folders/1uNY_rcOO5LrWibXRY6l2dvqSbK6xikJp?usp=sharing)\n* E2E aitodtoolkit. [[Folder]](aitodtoolkit)\n\n\u003c!-- * AI-TOD_wo_xview. [[BaiduDrive(Password:w2dy)]](https://pan.baidu.com/s/1AlhHIxpvkJ1-2ql9JdWqKg) [[OneDrive]](https://1drv.ms/u/s!Ao5UiAkIbGJ7xHCGhQe2tsU1Ut5i?e=SrUdYp) --\u003e\n\u003c!-- You can download the dataset on [Google Drive](https://drive.google.com/drive/folders/1mokzFtLCjyqalSEajYTUmyzXvOHAa4WX?usp=sharing) or [Baidu Drive](https://pan.baidu.com/s/1r2C_fBwQL4q2NRmDM3-RUw) (Password: 0ire). --\u003e\n\n## A Guide on Generating AI-TOD\n**Step 1:** Download the xView training set, AI-TOD without xview, and clone the aitodtoolkit.\n\n```\ngit clone https://github.com/jwwangchn/AI-TOD.git\n```\n\n**Step 2:** Organize the downloaded files in the following way.\n\n```\n├─aitod\n│  ├─annotations ## put the downloaded annotations of AI-TOD_wo_xview (.json)\n│  └─images ## unzip the downloaded AI-TOD_wo_xview image sets, put them (.png) in the corresponding folder\n│      ├─test ## directly put the images in it without extra folder\n│      ├─train \n│      ├─trainval \n│      └─val \n├─aitod_xview ## here are six files (.txt)\n├─xview\n│  ├─ori\n│  │   └─train_images ## unzip the downloaded xView training set images, put them (.tif) here\n│  └─xView_train.geojson ## the annotation file of xView training set\n└─generate_aitod_imgs.py ## end-to-end tool\n```\n\n**Step 3:** Install required packages.\n\n* Required environment\n1. Python 3.7\n2. [mmcv](https://github.com/open-mmlab/mmcv)\n\n* Install [wwtool](https://github.com/jwwangchn/wwtool)\n\n```\ngit clone https://github.com/jwwangchn/wwtool.git\ncd wwtool\npython setup.py develop\n```\n* Install other required packages\n\n```\ncd ..\ncd aitodtoolkit\npip install -r requirements.txt\n```\n\n**Step 4:** Run the E2E aitodtoolkit and get AI-TOD, it might take around an hour, then the full image sets of AI-TOD can be found in the **aitod** folder. And you can delete other files in other folders to avoid taking up too much space.\n\n```\npython generate_aitod_imgs.py\n```\n\n\n## Evaluation\nTraining, Validation and Testing sets are both publicly available now. We report the COCO style performance in the original paper, you can use the [cocoapi-aitod](https://github.com/jwwangchn/cocoapi-aitod) to evaluate the model performance.\n\n\n## Citation\n\nIf you use this dataset in your research, please consider citing these papers.\n\n```\n@inproceedings{AI-TOD_2020_ICPR,\n    title={Tiny Object Detection in Aerial Images},\n    author={Wang, Jinwang and Yang, Wen and Guo, Haowen and Zhang, Ruixiang and Xia, Gui-Song},\n    booktitle=ICPR,\n    pages={3791--3798},\n    year={2021},\n}\n```\n\n```\n@article{NWD_2021_arXiv,\n  title={A Normalized Gaussian Wasserstein Distance for Tiny Object Detection},\n  author={Wang, Jinwang and Xu, Chang and Yang, Wen and Yu, Lei},\n  journal={arXiv preprint arXiv:2110.13389},\n  year={2021}\n}\n```\n## Reference\n[xView Dataset](http://xviewdataset.org/)\n\n## License\n\nThe AI-TOD dataset is licensed under the Attribution-NonCommercial-ShareAlike 4.0 International ([CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/)). Thus AI-TOD dataset are freely available for academic purpose or individual reserach, but restricted for commercial use. Besides, the underlying codes are licensed under the MIT license.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjwwangchn%2FAI-TOD","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fjwwangchn%2FAI-TOD","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjwwangchn%2FAI-TOD/lists"}