Enhancing Rotated Object Detection in Remote Sensing with a Parallel Hybrid Attention Mechanism

The primary objective of this project is to develop an enhanced rotated object detection system for accurately identifying ships and airplanes in aerial and satellite images. It implements YOLO26_PHAM with a hybrid CBAM-SE attention module and YOLO11_SDA with Spiral Depthwise Attention to capture multi-scale, orientation-aware features for precise detection. Both models are trained and evaluated on a merged dataset, with metrics for validation. The system includes a Flask-based web interface for image upload and prediction display and compares enhanced models against baseline YOLO models to demonstrate the benefits of parallel hybrid attention. It is optimized for accuracy and computational efficiency, supports future extensions, and provides comprehensive documentation of methodology and results.