ROSA-TFormer: A Radar-Optical Sensor-Aware Temporal Transformer for Pinus sylvestris Plantation Classification in Northern Shaanxi Using GEE-Derived Sentinel-1/2 Time Series

2026-06-17 · Source: Computer Vision and Pattern Recognition · Field: Science & Research — Artificial Intelligence & Machine Learning, Environmental Science & Earth Systems · Depth: Expert, quick

Summary

ROSA-TFormer is a novel radar-optical sensor-aware temporal Transformer model designed for accurate classification of Pinus sylvestris var. mongolica plantations in northern Shaanxi. This model utilizes Sentinel-1/2 time-series data processed via Google Earth Engine, integrating distinct SAR and optical embedding branches, a sensor-aware gate, and temporal attention pooling to capture multi-source seasonal features. Experiments conducted on monthly and half-month point-level datasets demonstrate strong performance. Specifically, ROSA-TFormer achieved 99.67% overall accuracy, 99.56% macro F1, and 98.91% P. sylvestris F1 on the HalfMonth-dataBig dataset. Spatial block validation and ablation studies further confirm the effectiveness of its radar-optical temporal fusion and sensor-aware modeling approach for point-level plantation classification.

Key takeaway

For Research Scientists developing remote sensing models for ecological monitoring, ROSA-TFormer offers a robust architecture for Pinus sylvestris plantation classification. You should consider integrating radar-optical sensor-aware temporal Transformers into your workflows, especially when utilizing Sentinel-1/2 time-series data. This approach demonstrates high accuracy (e.g., 99.67% overall accuracy), suggesting a powerful method for improving afforestation quality assessment. However, remember that broader wall-to-wall validation is still necessary for full operational deployment.

Key insights

ROSA-TFormer effectively classifies Pinus sylvestris plantations by fusing radar and optical time-series data with sensor-aware temporal Transformers.

Principles

• Fusing radar and optical time series enhances plantation classification.
• Sensor-aware gating improves multi-source feature integration.
• Temporal attention pooling captures seasonal land cover changes.

Method

ROSA-TFormer integrates separate SAR and optical embedding branches, a sensor-aware gate, and temporal attention pooling to process Sentinel-1/2 time-series data for classification.

In practice

• Utilize GEE for Sentinel-1/2 time-series data generation.
• Implement sensor-aware gates for multi-modal remote sensing.
• Apply temporal Transformers for land cover change detection.

Topics

• ROSA-TFormer
• Pinus sylvestris Classification
• Radar-Optical Fusion
• Temporal Transformers
• Sentinel-1/2
• Google Earth Engine

Best for: AI Scientist, Research Scientist, Computer Vision Engineer

Related on AIssential

• FUSAR-GPT : A Spatiotemporal Feature-Embedded and Two-Stage Decoupled Visual Language Model for SAR Imagery — FUSAR-GPT enhances SAR VLM performance by integrating geospatial priors and decoupling training stages for robust interpretation.
• Landsat-Sentinel-2 Algal Bloom Mapping Using Vision Transformers: Model Description, Implementation, and Examples — Vision Transformers enable robust, high-resolution algal bloom mapping from harmonized Landsat-Sentinel-2 imagery, outperforming traditional methods under challenging conditions.
• deFOREST: Fusing Optical and Radar satellite data for Enhanced Sensing of Tree-loss — Fusing optical and SAR satellite data with a novel anomaly detection method enhances deforestation monitoring, especially in cloudy areas.
• An Open Dataset for the Acoustic Monitoring of Nocturnal Migratory Birds in Europe — Crowd-sourced acoustic data and a specialized deep learning model enhance nocturnal migratory bird monitoring.
• Heterogeneous LiDAR Early Fusion and Learned Re-Ranking Strategy for Robust Long-Term Place Recognition in Unstructured Environments — Heterogeneous LiDAR early fusion and learned re-ranking significantly boost place recognition in challenging unstructured agricultural environments.
• WILD-SAM: Phase-Aware Expert Adaptation of SAM for Landslide Detection in Wrapped InSAR Interferograms — WILD-SAM adapts SAM for landslide detection in InSAR data by integrating phase-aware spectral alignment and wavelet-guided prompt enhancement.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computer Vision and Pattern Recognition.