Field-Verified Crop Type Polygon Dataset for Rabi Season 2025 in Vijayapura District for Satellite-Based Crop Classification

Engineering notes

Engineering notes will be added by the aipentium editorial team.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为大语言模型、生成式AI、ChatGPT相关技术、计算机视觉、深度学习等高价值论文补充中文说明。

Original abstract

Dataset Description This dataset contains crop type sample polygons collected for Rabi season 2025 in Vijayapura District, intended for training machine learning models for crop type classification using satellite data. The dataset was collected through extensive field surveys conducted between December 2024 and April 2025. A total of approximately 4,000 crop sample polygons were mapped using field observations and geospatial tools. Each polygon represents a homogeneous agricultural field with a confirmed crop type during the Rabi growing season. The dataset was designed to support remote sensing–based crop classification studies using multi-temporal satellite imagery, particularly **Sentinel‑2 optical data and **Sentinel‑1 SAR data. The sample polygons serve as training data for supervised classification models, including deep learning architectures such as transformer-based temporal models. The spatial distribution of samples was designed to capture variability in agricultural practices, soil conditions, and crop phenology within the region. For model evaluation and transferability assessment, independent inference data were collected from Dharwad District, enabling testing of model performance across geographically distinct agricultural landscapes. This dataset represents one of the most extensive crop-type field datasets collected by an individual researcher in Karnataka, providing valuable ground reference data for geospatial artificial intelligence (GeoAI), agricultural monitoring, and remote sensing research. Data Collection Method Field samples were collected through ground surveys and field verification, where crop types were identified through direct observation during different stages of crop growth. Polygons were digitized to represent the full boundary of agricultural fields rather than single point observations, allowing more accurate extraction of satellite pixel information for machine learning workflows. Each polygon corresponds to a single crop type and represents a homogeneous agricultural parcel. Intended Use The dataset is suitable for: Crop type classification using satellite imagery Multi-temporal remote sensing analysis Training deep learning and machine learning models Sentinel-1 and Sentinel-2 data fusion studies Agricultural monitoring and GeoAI research Transfer learning experiments across regions Temporal Coverage December 2024 – April 2025(Rabi cropping season) Spatial Coverage Training Data: Vijayapura District Inference / Testing Region: Dharwad District Dataset Significance High-quality labeled field datasets are a major limitation for crop classification studies using satellite imagery. This dataset contributes a large, field-verified crop sample database from semi-arid agricultural landscapes of northern Karnataka and can support research in: precision agriculture food security monitoring satellite-based agricultural statistics machine learning for Earth observation.

Links and sources

Comments