Invariant Features for Global Crop Type Classification

2025-09-03

AI Summary

Global crop mapping is hindered by scarcity of ground truth samples and poor cross-regional generalization. To address this, we propose geographically invariant remote sensing feature representations and introduce CropGlobe—a pixel-level benchmark dataset covering eight countries and six major staple crops. We further design CropNet, a lightweight convolutional network, and a time-series–specific data augmentation strategy incorporating temporal shifting, scaling, and amplitude warping. Leveraging Sentinel-2 multi-temporal median and harmonic features alongside EMIT hyperspectral data, our experiments demonstrate that 2D median time-series features achieve superior performance in cross-national, intercontinental, and inter-hemispheric transfer tasks. The method attains robust classification with only limited labeled data, significantly enhancing scalability and enabling low-cost deployment for global-scale crop mapping.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Accurately obtaining crop type and its spatial distribution at a global scale is critical for food security, agricultural policy-making, and sustainable development. Remote sensing offers an efficient solution for large-scale crop classification, but the limited availability of reliable ground samples in many regions constrains applicability across geographic areas. To address performance declines under geospatial shifts, this study identifies remote sensing features that are invariant to geographic variation and proposes strategies to enhance cross-regional generalization. We construct CropGlobe, a global crop type dataset with 300,000 pixel-level samples from eight countries across five continents, covering six major food and industrial crops (corn, soybeans, rice, wheat, sugarcane, cotton). With broad geographic coverage, CropGlobe enables a systematic evaluation under cross-country, cross-continent, and cross-hemisphere transfer. We compare the transferability of temporal multi-spectral features (Sentinel-2-based 1D/2D median features and harmonic coefficients) and hyperspectral features (from EMIT). To improve generalization under spectral and phenological shifts, we design CropNet, a lightweight and robust CNN tailored for pixel-level crop classification, coupled with temporal data augmentation (time shift, time scale, and magnitude warping) that simulates realistic cross-regional phenology. Experiments show that 2D median temporal features from Sentinel-2 consistently exhibit the strongest invariance across all transfer scenarios, and augmentation further improves robustness, particularly when training data diversity is limited. Overall, the work identifies more invariant feature representations that enhance geographic transferability and suggests a promising path toward scalable, low-cost crop type applications across globally diverse regions.

Problem

• Identifying invariant remote sensing features for global crop classification
• Enhancing cross-regional generalization under geospatial shifts
• Developing robust models for scalable crop type mapping worldwide

Innovation

• Global crop dataset with 300K samples
• Lightweight CNN with temporal data augmentation
• 2D median temporal features from Sentinel-2