Computer vision and satellite imagery analysis for precision agriculture. Detects crop diseases, monitors soil health, predicts yields, and optimizes irrigation using drone imagery and multispectral data.

FedRSClip: Federated Learning for Remote Sensing Scene Classification Using Vision-Language Models (IEEE GRSM 2025)

海南部分红树林分类图，具体为花场湾和新盈港各两个年份的分类

Deep learning system for climate change analysis using satellite imagery and weather data. Predicts natural disasters, monitors deforestation, tracks glacier melting, and analyzes urban heat islands.

Data-curation levers (cross-dataset transfer and class-balanced sampling) for imbalanced high-resolution rural land-cover segmentation with FT-UNetFormer, and the label-quality ceiling that limits them.