Version : v1.0.0

Description

Fields of The World (FTW) is a comprehensive benchmark dataset designed to enhance the development of machine learning models for instance segmentation of agricultural field boundaries. This dataset aims to meet the growing need for accurate and scalable field boundary data for global agricultural monitoring and assessments.

Key Features:

1. Near-Global Coverage: FTW spans four continents—Europe, Africa, Asia, and South America—covering diverse agricultural landscapes across 24 countries. This extensive geographic coverage allows for the development of models that can generalize well to different agricultural practices and field types.

2. Large-Scale Dataset: With approximately 1.6 million parcel boundaries and over 70,000 samples, FTW is significantly larger than previously available datasets. Each sample includes instance and semantic segmentation masks paired with multi-date, multi-spectral Sentinel-2 satellite images, enabling detailed temporal and spectral analysis.

3. Multi-class Segmentation: The dataset provides masks for both instance segmentation and semantic segmentation with different classes, including:

   • Instance Segmentation Masks: To identify individual fields.
   • Semantic Segmentation Masks:
     • Two-class masks: Background and polygon (field).
     • Three-class masks: Background, polygon (field), and boundaries.

4. Spectral Richness: The dataset includes RGB (Red, Green, Blue) and NIR (Near-Infrared) spectral bands from Sentinel-2 images.

5. Temporal Richness: The dataset includes multi-date imagery to capture different stages of the growing season. Two images with distinct contrast differences were selected to represent these stages. To determine the date ranges for these images, the USDA Crop Calendar was initially referenced and then refined by selecting periods with minimal cloud cover and optimal contrast between the two images.

6. Comprehensive Data Splits: The dataset is carefully divided into training, validation, and test sets to ensure accurate evaluation of model performance. For each country, larger tiles are divided into smaller chips measuring 1536x1536 m². To prevent data leakage due to spatial autocorrelation, a blocked random splitting strategy is used. Chips are grouped into 3×3 blocks, with 80% allocated to training, 10% to validation, and 10% to testing.

7. Metadata and Documentation: The metadata and documentation provide crucial information to help users effectively interpret and utilize the dataset. It includes key details about the country of focus, temporal data collection windows, grid structures, and the year of collection.

   1. Country: The geographic region the dataset focuses on.
   2. Crop Types: Crop types used to filter the polygons in the dataset, with specific keywords used for filtering.
   3. Seasons: Date ranges defining the temporal windows for data collection.
   4. Year of Collection: The year when the polygon boundaries were captured.
   5. Grids: Larger grids from which smaller chips are derived, with some grids spatially separated to optimize area coverage.

---

Dataset Directory Structure

Fields of The World
├── README.md                      -> This File
├── ftw-sources.pmtiles            -> File to visualize the source field boundaries on https://fieldsofthe.world/map/
├── austria                        -> Country Folder
│   ├── label_masks                -> Labels Folder
│   │   ├── instance               -> Instance Segmented Masks (Label) (Masks in .tif Format)
│   │   ├── semantic_2class        -> Semantic Segmented Masks (Label) (Masks in .tif Format) -> Contains 2 Classes (0-Background, 1-Polygon)
│   │   └── semantic_3class        -> Semantic Segmented Masks (Label) (Masks in .tif Format) -> Contains 3 Classes (0-Background, 1-Polygon, 2-Boundaries)
│   ├── s2_images                  -> Images Folder (Contains image chips)
│   │   ├── window_a               -> Window A images (Images in .tif Format)
│   │   └── window_b               -> Window B images (Images in .tif Format)
│   ├── chips_austria.parquet      -> Chips file in geoparquet format, contains split details (Each chips belongs in one of Train/Val/Test split)
│   └── data_config_austria.json   -> Contains meta data about the bigger grids for the dataset, crop types, dates for temporal windows.
├── austria.zip                    -> Country Zip Folder, this contains all the files in the country directory.
└── checksum.md5                   -> Checksum MD5 file containing all the individual files checksum hashes. 
..... Continues for all the countries in the same format.

---

Dataset Information

Country	Year of Validity	Parcel Counts	Chips	Train Split	Validation Split	Test Split	Source Polygons	Source Data License
Austria	2021	196101	6686	5304	637	745	Link	CC-BY-4.0
Belgium	2021	63431	1941	1554	189	198	Link	No restrictions on public access
Brazil	2020	1854	1607	1289	130	188	Link	CC-BY-4.0
Cambodia	2021	318088	344	274	36	34	Link	CC-BY-4.0
Corsica	2021	5360	2472	1974	240	258	Link	CC-BY-2.0
Croatia	2023	157481	3482	2778	351	353	Link	Open Data
Denmark	2021	37677	3560	2868	360	332	Link	CC0-1.0
Estonia	2021	26695	6713	5348	681	684	Link	CC-3.0
Finland	2021	57323	5665	4527	550	588	Link	CC-BY-4.0
France	2020	55342	3744	2988	360	396	Link	Open Licence
Germany	2018/2019	4598	686	306	30	350	Link	DL-DE/BY-2-0
India	2016	10013	2002*	1281	300	399	Link	CC-BY-4.0
Kenya	2022	874	391	316	20	55	Link	GPL-2.0-or-later
Latvia	2021	44964	6938	5529	668	741	Link	CC-BY-NC-4.0
Lithuania	2021	61424	5258	4208	522	528	Link	Non-commercial use only
Luxembourg	2022	29018	808	643	81	84	Link	CC0-1.0
Netherlands	2022	43169	3879	3110	381	388	Link	CC0-1.0
Portugal	2021	5040	86	64	12	10	Link	CC-BY-NC-4.0
Rwanda	2021	1532	70	57	6	7	Link	CC-BY-4.0
Slovakia	2021	14242	4073	3275	390	408	Link	CC0-1.0
Slovenia	2021	67488	2177	1733	216	228	Link	CC-BY-4.0
South Africa	2018	6568	747	590	72	85	Link	CC-BY-NC-SA-4.0
Spain	2020	258465	2440	2019	202	219	Link	CC-BY-4.0
Sweden	2021	39718	4760	3802	442	516	Link	No restrictions on public access
Vietnam	2021	120913	288	229	36	23	Link	CC-BY-4.0

*India has a total of 2,002 chips available. Of these, 22 chips are marked as 'none' for the split column as per the original data curator. Thus, 1,980 chips have been used in the train/validation/test splits in India.