Abstract:Agricultural landscapes are quite complex, especially in the Global South where fields are smaller, and agricultural practices are more varied. In this paper we report on our progress in digitizing the agricultural landscape (natural and man-made) in our study region of India. We use high resolution imagery and a UNet style segmentation model to generate the first of its kind national-scale multi-class panoptic segmentation output. Through this work we have been able to identify individual fields across 151.7M hectares, and delineating key features such as water resources and vegetation. We share how this output was validated by our team and externally by downstream users, including some sample use cases that can lead to targeted data driven decision making. We believe this dataset will contribute towards digitizing agriculture by generating the foundational baselayer.

What problem does this paper attempt to address?

### What problems does this paper attempt to solve? This paper aims to address the complex challenges in Agricultural Landscape Understanding (ALU), especially the digitization of agricultural landscapes within a country like India. Specifically, the paper attempts to achieve accurate identification and classification of various features in the agricultural landscape through high - resolution satellite imagery and deep - learning techniques, thereby providing data support for sustainable agricultural management and policy - making. #### Main problems: 1. **Complexity of the agricultural landscape**: Especially in the Global South, the small size of farmland and the diverse agricultural practices make accurate mapping and monitoring difficult. 2. **Lack of detailed agricultural landscape data**: The existing agricultural landscape data are usually not detailed enough to support precise resource allocation and decision - making. 3. **Needs of small - scale farmers**: Small - scale farmers (with a cultivated area of less than 2 hectares) account for 86.2% of India's agricultural labor force and manage about half of the arable land, but these farms are often not recorded and analyzed in detail. 4. **Identification of multi - class features**: It is necessary not only to identify farmland boundaries, but also to identify key features such as water bodies and trees in order to fully understand the agricultural ecosystem. 5. **Data verification and reliability**: To ensure the reliability and practical application value of the model output, strict on - site verification is required. #### Solutions: - **Multi - class panoptic segmentation**: Using high - resolution satellite imagery based on RGB channels, ALU is modeled as a multi - class panoptic segmentation problem. Panoptic segmentation can not only divide the image into semantically meaningful regions, but also identify individual instances in these regions (such as different fields, independent water bodies, etc.). - **Fine - grained mapping at the national level**: For the first time, fine - grained mapping of agricultural land use has been achieved nationwide, especially with detailed identification of small - scale farmers' farmland. - **Strict on - site verification**: Through cooperation with external partners, detailed on - site verification has been carried out to ensure the accuracy and practicality of the model output. Through these methods, the paper provides an in - depth understanding of the agricultural landscape and lays the foundation for subsequent data - driven decision - making and policy - making.

Agricultural Landscape Understanding At Country-Scale

Agriculture-Vision: A Large Aerial Image Database for Agricultural Pattern Analysis

Generating Land Cover Maps in Semi-arid Regions Based on a 3D Semantic Segmentation Architecture Using Multi-temporal Sentinel-2 Satellite Images: A Case Study of Ludhiana District in Punjab, India

Unlocking Large-Scale Crop Field Delineation in Smallholder Farming Systems with Transfer Learning and Weak Supervision

Investigating the Segment Anything Foundation Model for Mapping Smallholder Agriculture Field Boundaries Without Training Labels

Towards Space-to-Ground Data Availability for Agriculture Monitoring

Detecting The Boundaries Of Urban Areas In India: A Dataset For Pixel-Based Image Classification In Google Earth Engine

Fields of The World: A Machine Learning Benchmark Dataset For Global Agricultural Field Boundary Segmentation

High-Resolution Satellite Imagery for Modeling the Impact of Aridification on Crop Production

Advancing crop classification in smallholder agriculture: A multifaceted approach combining frequency-domain image co-registration, transformer-based parcel segmentation, and Bi-LSTM for crop classification

A generalized framework for agricultural field delineation from high-resolution satellite imageries

A fusion approach using GIS, green area detection, weather API and GPT for satellite image based fertile land discovery and crop suitability

Ultra-high-resolution hyperspectral imagery datasets for precision agriculture applications

Machine learning based object-level crop classification of PlanetScope data at South India Basin

In-situ and airborne hyperspectral data for detecting agricultural activities in a dense forest landscape

Multi-Region Transfer Learning for Segmentation of Crop Field Boundaries in Satellite Images with Limited Labels

Extended Agriculture-Vision: An Extension of a Large Aerial Image Dataset for Agricultural Pattern Analysis

A Sentinel-2 multi-year, multi-country benchmark dataset for crop classification and segmentation with deep learning

A hybrid UNet based approach for crop classification using Sentinel-1B synthetic aperture radar images

An Artificial Intelligence Dataset for Solar Energy Locations in India

Application of Multi-Temporal and Multisource Satellite Imagery in the Study of Irrigated Landscapes in Arid Climates