Upload folder using huggingface_hub

README.md

@@ -1,89 +1,67 @@
 ---
-license: cc-by-nc-4.0
+language: en
+license: unknown
+task_categories:
+- object-detection
+paperswithcode_id: FAIR1M
+pretty_name: FAIR1M
+tags:
+- remote-sensing
+- earth-observation
+- geospatial
+- satellite-imagery
+- object-detection
 ---
 
-## Domains
-| 序号 | 无人机 Domain      | 领域定位                         | 典型检测目标                                                                                                                    | 典型应用                                                                                                                      | 核心输出 / 特点                                                   |
-| -- | ---------------- | ---------------------------- | ------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------- |
-| 1  | **农林牧业与生态监测**    | 检测自然环境中的生物目标及农业生产对象          | 农作物、果实、花朵、植株<br>缺苗、倒伏作物、杂草<br>牛、羊等牲畜<br>野生动物、鸟类<br>病树、枯死树、倒伏树木<br>农业机械、作业人员                                               | 作物株数和果实数量统计<br>缺苗、杂草和倒伏区域发现<br>牲畜自动盘点和走失牲畜搜索<br>野生动物检测、跟踪和数量调查<br>森林病树、枯死树和风倒木识别<br>农林区域非法人员或车辆检测                         | 目标类别<br>目标数量<br>目标位置<br>目标密度<br>变化趋势                        |
-| 2  | **能源、工业与基础设施巡检** | 将电力、油气、工厂、桥梁、道路等统一归入资产和设施巡检  | 输电杆塔、导线、绝缘子、金具<br>鸟巢、异物、悬挂物<br>光伏组件、热斑组件<br>风机叶片、裂纹及表面损伤<br>管道、阀门、储罐、烟囱<br>桥梁构件、道路设施、铁路设备<br>裂缝、锈蚀、破损、脱落<br>设备热点、泄漏点和异常部件 | 电力线路异物和缺陷检测<br>绝缘子、连接件和杆塔部件检测<br>光伏板热斑及故障组件定位<br>风机叶片损伤检测<br>油气管线和工业设备异常检测<br>桥梁、道路、铁路设施缺陷识别<br>设备缺失、错位或损坏检测              | **部件检测：**定位设备及零部件<br>**缺陷检测：**识别裂纹、腐蚀、破损、过热等异常              |
-| 3  | **建筑、施工与矿山作业**   | 将施工现场、露天矿、采石场和大型工程现场视为复杂作业区域 | 施工人员<br>安全帽、反光衣、安全绳等防护装备<br>挖掘机、吊车、卡车、推土机<br>建筑材料和堆料<br>临时建筑、围栏和危险区域<br>裂缝、坍塌、落石<br>非法进入人员和车辆<br>烟雾、明火和积水                 | 施工人员和作业车辆统计<br>安全帽、反光衣等PPE检测<br>人员进入危险区域检测<br>起重设备周边人员检测<br>矿区车辆及机械设备调度监测<br>边坡落石、裂缝和坍塌检测<br>非法采矿、非法占地目标发现<br>施工现场火灾和烟雾检测 | 判断现场有哪些目标<br>确定目标所在位置<br>判断人员、车辆或设备是否违反安全规则                 |
-| 4  | **公共安全、应急与安防**   | 将消防、搜救、灾害响应、边境和园区安防整合为安全应用领域 | 人员、失踪者、受困人员<br>车辆、船只<br>明火、烟雾和高温热点<br>聚集人群<br>越界人员<br>可疑车辆或可疑物体<br>灾区生命目标<br>夜间移动目标                                       | 山区、森林和水域搜救<br>红外人体检测<br>火点、烟雾和复燃点检测<br>洪水、地震等灾害后的人员和车辆搜索<br>园区、边境及重要设施周界巡逻<br>人群聚集和异常活动检测<br>夜间入侵人员检测<br>危险区域非法进入检测       | 红外相机可用于夜间人员检测<br>烟雾遮挡环境中的热目标发现<br>森林或复杂地形中的生命目标搜索<br>火灾热点定位 |
-| 5  | **交通与智慧城市**      | 关注道路、城市空间及交通系统中的动态目标检测       | 小汽车、卡车、公交车<br>摩托车、自行车<br>行人<br>停放车辆<br>交通事故车辆<br>道路障碍物<br>抛洒物、落石<br>占道施工设备                                                | 车辆检测、分类和计数<br>车流量和道路拥堵分析<br>车辆轨迹和速度分析<br>违停、逆行和异常停车检测<br>交通事故自动发现<br>道路障碍物和抛洒物检测<br>大型活动人群数量统计<br>停车场车位占用检测               | 目标检测负责发现目标<br>多目标跟踪负责持续识别同一目标<br>可进一步计算车速、轨迹、停留时间和交通流量      |
-| 6  | **海洋���水域与环境执法**   | 将海岸、河流、湖泊、港口及海上监管统一归入水域监测    | 船舶、渔船和小型快艇<br>漂浮人员和救生设备<br>海洋动物<br>漂浮垃圾<br>水面油污和异常漂浮物<br>非法采砂船<br>非法捕捞船只<br>河道及岸边非法排放设施                                   | 船舶检测、分类和数量统计<br>非法捕捞及越界船只检测<br>水上搜救和落水人员检测<br>海洋动物检测和数量调查<br>河流、湖泊漂浮垃圾检测<br>水面异常目标和污染物发现<br>港口和海岸线安防巡逻<br>非法采砂、非法排放行为发现   | 重点解决水面反光、波浪干扰和远距离小目标问题<br>可结合长焦、红外和视频跟踪提升检测效果               |
+# FAIR1M
 
+<!-- Dataset thumbnail -->
+![FAIR1M](./thumbnail.jpg)
 
-| 序号 | 卫星 Domain    | 典型检测目标                                                                                                                                                                                                                                               | 多光谱数据与公开数据集依据                                                                                                                                                                                                                                                                                         | 典型任务与输出                                                                                     |
-| -- | ---------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------- |
-| 1  | **海事、航空与交通目标监测** | **海事目标：**船舶、渔船、大型货船、海上平台及港口目标<br><br>**航空目标：**飞机、直升机及机场地面航空器<br><br>**飞机细粒度类型：**Boeing 737、747、777、787，Airbus A220、A321、A330、A350，ARJ21、C919及其他飞机类型<br><br>**飞机属性：**机身长度、翼展、机翼形状、机翼位置、动力类型、发动机数量、垂尾数量及飞机用途<br><br>**交通设施：**机场、跑道、停机坪、港口、船厂、道路桥梁和铁路桥梁 | **S2-SHIPS**使用Sentinel-2多光谱影像开展船舶检测，说明船舶是原生多光谱目标检测中相对成熟的方向。([MDPI][1])<br><br>**fMoW-full**提供4波段和8波段可见光—近红外影像，可支持机场、港口、船厂和交通设施等目标级识别。([GitHub][2])<br><br>**FAIR1M**提供旋转框细粒度遥感目标检测，明确要求模型不仅检测“飞机”，还要识别Airbus A350、Boeing 747等具体类型。([ar5iv][3])<br><br>**RarePlanes**提供飞机检测以及10类细粒度结构属性标注。([arXiv][4]) | 船舶和飞机定位、分类、计数<br>飞机具体型号识别<br>飞机结构属性与用途分类<br>飞机停放位置及朝向检测<br>船舶尺寸、方向和航行状态估计<br>机场、港口及交通枢纽活动监测 |
-| 2  | **城市建筑与公共基础设施**  | 建筑物、住宅区和大型厂房<br>施工场地及临时建筑<br>机场和公共服务设施<br>道路、桥梁、水坝和交通节点<br>受损、倒塌或被洪水影响的建筑物                                                                                                                                                                           | **fMoW-full**的多光谱数据覆盖住宅、施工场地、医院、学校、公共机构、桥梁和水坝等设施类别。([GitHub][2])<br><br>**SatlasPretrain**融合Sentinel-2与高分辨率航片，提供建筑、机场、道路等多种点、线和多边形标注。([CVF Open Access][5])                                                                                                                                           | 建筑物和基础设施定位<br>设施数量及空间分布统计<br>新增建筑和施工活动发现<br>公共设施识别<br>灾前灾后建筑目标对比                            |
-| 3  | **能源、工业与资源设施**   | 光伏电站和光伏阵列<br>风力发电机及风电场<br>储油罐、储罐区和炼化设施<br>发电厂、变电站及工业园区<br>海上油气平台<br>矿区、采石场和资源开发设施<br>水处理和垃圾处理设施                                                                                                                                                      | **fMoW-full**包含光伏电站、风电场、发电厂、变电站、油气设施、储罐、露天矿和水处理设施等多光谱目标类别。([GitHub][2])<br><br>**SatlasPretrain**包含风机点目标、光伏场区域目标、建筑和其他基础设施标注，并使用Sentinel-2等卫星数据。([CVF Open Access][5])                                                                                                                                | 能源资产检测和计数<br>工业设施定位与分类<br>风机、光伏场和海上平台普查<br>储罐及大型设备统计<br>资源开发活动和设施变化监测                       |
+<!-- Provide a quick summary of the dataset. -->
+The FAIR1M dataset is a fine-grained object recognition and detection dataset that focuses on high-resolution (0.3-0.8m) RGB images taken by the Gaogen (GF) satellites and extracted from Google Earth. It consists of a collection of 15,000 high-resolution images that cover various objects and scenes. The dataset provides annotations in the form of rotated bounding boxes for objects belonging to 5 main categories (ships, vehicles, airplanes, courts, and roads), further divided into 37 sub-categories.
+- **Paper:** https://arxiv.org/abs/2103.05569
+- **Homepage:** https://www.gaofen-challenge.com/benchmark
 
-[1]: https://www.mdpi.com/2072-4292/13/21/4255?utm_source=chatgpt.com "Ship Detection in Sentinel 2 Multi-Spectral Images with Self ..."
-[2]: https://github.com/fMoW/dataset?utm_source=chatgpt.com "fMoW/dataset"
-[3]: https://ar5iv.org/pdf/2103.05569 "[2103.05569] FAIR1M: A Benchmark Dataset for Fine-grained Object Recognition in High-Resolution Remote Sensing Imagery"
-[4]: https://arxiv.org/abs/2006.02963 "[2006.02963] RarePlanes: Synthetic Data Takes Flight"
-[5]: https://openaccess.thecvf.com/content/ICCV2023/papers/Bastani_SatlasPretrain_A_Large-Scale_Dataset_for_Remote_Sensing_Image_Understanding_ICCV_2023_paper.pdf?utm_source=chatgpt.com "SatlasPretrain: A Large-Scale Dataset for Remote Sensing ..."
+## Description
 
+<!-- Provide a longer summary of what this dataset is. -->
 
+FAIR1M is a part of the ISPRS Benchmark on Object Detection in High-Resolution Satellite Images. Please note that, as of now, only a portion of the training dataset (1,732/15,000 images) has been released for the challenge.
 
-## Dataset Structure
-| Domain | Dataset | Task | Output Format | Path | Source | Modality | Included|
-|---------|---------|-------------|---------------|------|------|------|------|
-| 1-1 | AgroVG | Agricultural Visual Grounding | Box  | `ObjEarth-Data/AgroVG/t1` | [here](https://huggingface.co/datasets/sauryrs/AgroVG) | RGB  | ✅ |
-| 1-1 | AgroMind | agricultural remote sensing benchmark(13 tasks) | ***  | `ObjEarth-Data/AgroMind/` | [here](https://huggingface.co/datasets/AgroMind/AgroMind) | RGB  | ✅ |
-| 1-1 | gRef-CW | Generalised Visual Grounding in Agriculture | Box  | `ObjEarth-Data/gRef-CW/` -  `ObjEarth-Bench/gRef-CW/`  | [here](https://huggingface.co/datasets/Mhaghighat98/gRef-CW) | RGB  | ✅ |
-| 1-1 | DroneBird | Bird detection, counting, and tracking | Box  | `ObjEarth-Data/DroneBird/` | [here](https://visdrone.org/visdrone/data-base/dronebird) | RGB  | ✅ |
-| 1-1 | AnimalDrone | Animal detection and counting | Box  | `ObjEarth-Data/AnimalDrone/` | [here](https://visdrone.org/visdrone/data-base/animaldrone) | RGB  | ✅ |
-| 1-1 | TinyTrees | Tree counting | point  | `ObjEarth-Data/tinytrees/` | [here](https://github.com/dgominski/treematch) |  satellite imagery | ✅ |
-| 1-2 | STN-PLAD |Power Line Assets Detection | Box  | `ObjEarth-Data/STN-PLAD` | [here](https://github.com/andreluizbvs/PLAD) |  RGB | ✅ |
-| 1-2 | InsPLAD | Power Line Asset Inspection | Box  | `ObjEarth-Data/InsPLAD` | [here](https://github.com/andreluizbvs/InsPLAD) |  RGB | ✅ |
-| 1-2 | WTBD | Wind Turbine Blade Detection | Box  | `ObjEarth-Data/WTBD` | [here](https://springernature.figshare.com/articles/dataset/Multiclass_Dataset_for_Intelligent_Detection_of_Wind_Turbine_Blade_Defects_Using_Drone_Imagery/30210175) |  RGB | ✅ |
-| 1-2 | TTPLA | Detection and Segmentation of Transmission Towers and Power Lines | Box+Mask  | `ObjEarth-Data/TTPLA` | [here](https://github.com/R3ab/ttpla_dataset) |  RGB | ✅ |
-| 1-3 | UAVConstructionSite | UAV Construction Site Image Captioning | None  | `ObjEarth-Data/UAVConstructionSite` | [here](https://data.mendeley.com/datasets/4h68fmktwh/1) |  RGB | ✅ |
-| 1-4 | uav_dataset | Person Detection Covered with Heavy Snow (Search and Rescue) | Box  | `` | [here](https://www.kaggle.com/datasets/imbalet/uav-dataset) |  RGB | ⏳ |
-| 1-4 | FLAME2-DT | Wildland Fire Detection | Box  | `ObjEarth-Data/FLAME2-DT/` | [here](https://ieee-dataport.org/documents/flame2-dt#files)  | RGB+Infrared |✅ |
-| 1-4 | SARD   | Person Detection in Search and Rescue Scenarios in Drone | Box  | `ObjEarth-Data/SARD-fixed` | [here](https://ieee-dataport.org/documents/search-and-rescue-image-dataset-person-detection-sard)  | RGB | ✅|
-| 1-4 | RescueNet  | Post Disaster UAV Dataset for Semantic Segmentation | Mask  | `ObjEarth-Data/RescueNet` | [here](https://github.com/BinaLab/RescueNet-A-High-Resolution-Post-Disaster-UAV-Dataset-for-Semantic-Segmentation) |  RGB | ⏳ |
-| 1-5 | UAV-SEG-2025 | UAV Pavement Distress Detection |  Box | `ObjEarth-Data/UAV-SEG-2025` | [here](https://www.kaggle.com/datasets/patelravi07/enhanced-uav-pdd) | RGB  | ✅ |
-| 1-5 | UAVVaste  | UAV Trash and Litter Detection |  Box | `ObjEarth-Data/UAVVaste/` | [here](https://www.kaggle.com/datasets/vexxingbanana/uavvaste-dataset) | RGB  | ✅ |
-| 1-5 | RefDrone | Drone Object Referring | Box | `ObjEarth-Data/VisDrone2019_DET/` - `ObjEarth-Bench/RefDrone/`| [here](https://github.com/sunzc-sunny/refdrone) | RGB | ✅ |
-| 1-5 | VisDrone2019_DET | Drone Object Detection | Box | `ObjEarth-Data/VisDrone2019_DET/` |    | RGB | ✅ |
-| 1-5 | HazyDet | Drone Object Detection in Hazy | Box  | `ObjEarth-Data/HazyDet/` |   | RGB  |✅ |
-| 1-6 | AFO | Aerial Floating Object Detection (Maritime Search and Rescue) | Box  | `ObjEarth-Data/AFO` | [here](https://www.kaggle.com/datasets/jangsienicajzkowy/afo-aerial-dataset-of-floating-objects) |  RGB | ✅ |
-| 1-6 | FloPWD  | Floating Plastic Waste Segmentation | Mask  | `ObjEarth-Data/FloPWD` | [here](https://data.mendeley.com/datasets/znxjncgjkc/2) | RGB  | ✅ |
-| 2-1 | VHRShips | Ship Detection | Box  | `ObjEarth-Data/VHRShips` | [here](https://github.com/radres333/VHRShips) | Optical satellite images  | ✅ |
-| 2-1 | HRSID |  Ship Detection | Box  | `ObjEarth-Data/HRSID` | [here](https://github.com/chaozhong2010/HRSID) | SAR  | ✅ |
-| 2-1 | MAR20 | Military Aircraft Detection | Box  | `ObjEarth-Data/MAR20` | [here](https://1drv.ms/u/s!AmgKYzARBl5ceGUKiVsRzfxZa_4?e=K21Spg) |  RGB | ✅ |
-| 2-3 | OPMOD | Open Pit Mine Object Detection |  Box | `ObjEarth-Data/OPMOD` | [here](https://figshare.com/articles/dataset/Open_Pit_Mine_Object_Detection_Dataset/27300960) | Optical satellite images | ✅ |
-| 2-3 | MARIDA | Marine Debris Segmentation  | Mask  | `ObjEarth-Data/MARIDA` | [here](https://zenodo.org/records/5151941) | Multispectral Sentinel-2  | ✅ |
-| | DVGBench | UAV Visual Grounding | Box | `ObjEarth-Data/DVG_Bench/images/` -  `ObjEarth-Bench/DVGBench/` |  [here](https://github.com/VisionXLab/DVGBench) | RGB | ✅ |
-| | TriModalDet | UAV Object Detection |  Box | `ObjEarth-Data/TriModalDet/` | [here](https://github.com/radlab-sketch/trimodal-uav-det) |  RGB–Thermal–Event camera | ✅|
-| 1-5 | AU-AIR | UAV Object Detection (Traffic Surveillance)| Box | `ObjEarth-Data/AU-AIR/images/` |  | RGB | ⏳ |
-| 1-3 | CIS | Construction Instance Segmentation | Box+Mask  | `` | [here](https://github.com/XZ-YAN/CIS-Dataset) |  RGB | * |
-| 1-3 | ELDOR  | Illegal Gold Mining |  Mask | `` | [here](https://huggingface.co/datasets/IRSC/ELDOR)  | RGB  | *  |
-| 1-4 | TinyPerson | Person Detection for Maritime Rescue | Box  | `` | [here](https://github.com/ucas-vg/PointTinyBenchmark) | RGB  |  |
-| 1-4 | FASDD_UAV | Flame and Smoke Detection | Box  | `` |   [here](https://www.scidb.cn/detail?dataSetId=ce9c9400b44148e1b0a749f5c3eb0bda) |  RGB | * |
-| 1-6 | MOBDrone | Man OverBoard Rescue Detection | Box  | `` | [here](https://aimh.isti.cnr.it/dataset/MOBDrone/) |  RGB |  |
-| 1-6 | SeaDronesSee | Human Detection (Maritime Search and Rescue) | Box  | `` | [here](https://github.com/Ben93kie/SeaDronesSee) |  RGB |  |
-| 1-6 | AVD | Aerial Vessels Detection | Box  | `` | [here](https://zenodo.org/records/7076145) |  RGB | * |
-| 1-6 | UAVPorpoises  | Drone-based Porpoise Tracking | Box  | `` | [here](https://putvision.github.io/UAVPorpoises/) | Video  | * |
-| 2-1 | ShipRSImagaeNet | Ship Detection | Box  | `` | [here](https://github.com/zzndream/ShipRSImageNet) |  Optical satellite images | * |
-| 2-1 | xView3-SAR | Ship Detection |  Box | `` | [here](https://iuu.xview.us/) |  SAR | * |
-| 2-1 | xView  | Object Detection for Overhead Imagery | Box  | `` | [here](https://xviewdataset.org/) |  RGB | * |
-| 2-1 | FAIR1M (1,732 images been released)  | Object Detection in High-Resolution Satellite Images|  Box | `` | [here](https://huggingface.co/datasets/blanchon/FAIR1M) | RGB  | * |
-| 2-1 | DIOR | Object Detection | Box  | `` | [here](https://drive.google.com/drive/folders/1UdlgHk49iu6WpcJ5467iT-UqNPpx__CC) | RGB  |  * |
-| 2-1 | DOTA | Object Detection in Aerial Images | Box  | `` | [here](https://captain-whu.github.io/DOTA/dataset) | RGB  |  * |
-| 2-1 | RarePlanes  | Aircraft Detection |  Box | `` | [here](https://www.iqt.org/library/the-rareplanes-dataset) |  Optical satellite images | * |
-| 2-1 | SADD  | SAR Aircraft Detection |  Box | `` | [here](https://github.com/hust-rslab/SAR-aircraft-data) |  SAR | * |
-| 2-1 | LEVIR-Ship | Tiny Ship Detection |  Box | `` | [here](https://github.com/WindVChen/LEVIR-Ship) | RGB  | * |
-| 2-1 | fMoW-full (3.5TB) | Object Detection | Box  | `` | [here](https://github.com/fMoW/dataset) |  Multispectral Imagery | * |
-| 2-1 | fMoW-rgb (200GB) | Object Detection | Box  | `` | [here](https://github.com/fMoW/dataset) |  RGB | * |
-| 2-1 | SatlasPretrain | large-scale pre-training dataset for remote sensing image understanding  | points, polygons, polylines, properties, segmentation labels, regression labels, and classification labels  | `` | [here](https://huggingface.co/allenai/satlas-pretrain) | RGB  | * |
-| 2-3 | FASDD_RS | Flame and Smoke Detection | Box  | `` |   [here](https://www.scidb.cn/detail?dataSetId=ce9c9400b44148e1b0a749f5c3eb0bda) |  Optical satellite images | * |
-| 2-3 | MADOS | Marine Debris and Oil Spill Segmentation | Mask  | `` | [here](https://zenodo.org/records/10664073) | Multispectral Sentinel-2  | * |
-| 2-1 |  |  |   | `` | [here]() |   | ❌ |
-| 2-1 |  |  |   | `` | [here]() |   | ❌ |
-| |  |  |   | `` |  |   | ❌ |
+- **1 million object instances**
+- **Number of Samples**: 15000
+- **Bands**: 3 (RGB)
+- **Image Size**: 1024x1024
+- **Image Resolution**: 0.3–0.8m
+- **Land Cover Classes**: 37
+- **Classes**: 5 object categories, 37 object sub-categories.
+- **Scene Categories**: Passenger Ship, Motorboat, Fishing Boat, Tugboat, other-ship, Engineering Ship, Liquid Cargo Ship, Dry Cargo Ship, Warship, Small Car, Bus, Cargo Truck, Dump Truck, other-vehicle, Van, Trailer, Tractor, Excavator, Truck Tractor, Boeing737, Boeing747, Boeing777, Boeing787, ARJ21, C919, A220, A321, A330, A350, other-airplane, Baseball Field, Basketball Court, Football Field, Tennis Court, Roundabout, Intersection, Bridge
+- **Source**: Gaofen/Google Earth
 
 
+## Usage
+
+To use this dataset, simply use `datasets.load_dataset("blanchon/FAIR1M")`.
+<!-- Provide any additional information on how to use this dataset. -->
+```python
+from datasets import load_dataset
+FAIR1M = load_dataset("blanchon/FAIR1M")
+```
+
+## Citation 
+
+<!-- If there is a paper or blog post introducing the dataset, the APA and Bibtex information for that should go in this section. -->
+If you use the FAIR1M  dataset in your research, please consider citing the following publication:
+
+
+```bibtex
+@article{sun2021fair1m,
+    title     = {FAIR1M: A Benchmark Dataset for Fine-grained Object Recognition in High-Resolution Remote Sensing Imagery},
+    author    = {Xian Sun and Peijin Wang and Zhiyuan Yan and F. Xu and Ruiping Wang and W. Diao and Jin Chen and Jihao Li and Yingchao Feng and Tao Xu and M. Weinmann and S. Hinz and Cheng Wang and K. Fu},
+    journal   = {Isprs Journal of Photogrammetry and Remote Sensing},
+    year      = {2021},
+    doi       = {10.1016/j.isprsjprs.2021.12.004},
+    bibSource = {Semantic Scholar https://www.semanticscholar.org/paper/6d3c2dc63ff0deec10f60e5a515c93af4f8676f2}
+}
+```