GTPBD: A Fine-Grained Global Terraced Parcel and Boundary Dataset

2025-07-19

AI Summary

Existing agricultural field extraction methods are primarily designed for medium-resolution imagery or regular plain terrains, failing to capture the fine-grained structural complexity of terraced landscapes—thus limiting precision agriculture applications. To address this, we introduce GTPBD, the first global high-resolution, fine-grained terraced-field parcel and boundary dataset, covering China’s seven major geographical regions and diverse climatic zones, with over 200,000 manually annotated terraced parcels. We propose a novel three-tier annotation scheme—guided by terrain diversity, irregular geometry, and cross-domain stylistic variation—that encompasses pixel-level boundaries/masks and parcel-level semantic labels. Further, we establish a multi-dimensional evaluation framework integrating both pixel-level and object-level metrics. Benchmarking across eight segmentation, four edge-detection, three parcel-extraction, and five unsupervised domain adaptation methods demonstrates significantly improved generalization on complex terraces, thereby bridging critical gaps in fine-grained agricultural remote sensing and cross-scenario knowledge transfer research.

Technology Category

• Computer Vision: Segmentation
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Web Mining and Content Analysis: Bridging structured and unstructured data

Abstract

Agricultural parcels serve as basic units for conducting agricultural practices and applications, which is vital for land ownership registration, food security assessment, soil erosion monitoring, etc. However, existing agriculture parcel extraction studies only focus on mid-resolution mapping or regular plain farmlands while lacking representation of complex terraced terrains due to the demands of precision agriculture.In this paper, we introduce a more fine-grained terraced parcel dataset named GTPBD (Global Terraced Parcel and Boundary Dataset), which is the first fine-grained dataset covering major worldwide terraced regions with more than 200,000 complex terraced parcels with manual annotation. GTPBD comprises 47,537 high-resolution images with three-level labels, including pixel-level boundary labels, mask labels, and parcel labels. It covers seven major geographic zones in China and transcontinental climatic regions around the world.Compared to the existing datasets, the GTPBD dataset brings considerable challenges due to the: (1) terrain diversity; (2) complex and irregular parcel objects; and (3) multiple domain styles. Our proposed GTPBD dataset is suitable for four different tasks, including semantic segmentation, edge detection, terraced parcel extraction, and unsupervised domain adaptation (UDA) tasks.Accordingly, we benchmark the GTPBD dataset on eight semantic segmentation methods, four edge extraction methods, three parcel extraction methods, and five UDA methods, along with a multi-dimensional evaluation framework integrating pixel-level and object-level metrics. GTPBD fills a critical gap in terraced remote sensing research, providing a basic infrastructure for fine-grained agricultural terrain analysis and cross-scenario knowledge transfer.

Problem

• Lack of fine-grained terraced parcel datasets for precision agriculture
• Challenges in mapping diverse terrains and irregular parcel shapes
• Need for multi-task dataset supporting segmentation and domain adaptation

Innovation

• First fine-grained global terraced parcel dataset
• High-resolution images with three-level labels
• Benchmarked on multiple tasks and methods