Real-Time Object Detection in High-Resolution Videos using Pan-Sharpening and CNNs*

Due to rising computing demands, high-resolution video object recognition is difficult. Traditional methods struggle to achieve real-time detection accuracy. This research study presents a two-step approach that uses pan-sharpening and CNNs to recognize objects in high-resolution films in real time. Pan-sharpening, a well-known image improvement technique, is the initial phase of the proposed framework. Fusing these photos improves video frame spatial resolution, revealing small details and subtle object attributes. Pan-sharpening improves object recognition by sharpening even small, distant objects. In the second step, a modern CNN architecture detects objects in pan-sharpened video frames in real time. CNNs understand complicated patterns and features from data effectively. A big dataset trains the CNN to accurately recognize and localize items. Real-time inference optimizes frame processing without sacrificing detection efficiency. High-resolution video datasets with a variety of objects, settings, and lighting conditions are used to evaluate the suggested technique. The framework outperforms typical object detection approaches in real-world experiments. Pan-sharpening and CNNs improve object detection accuracy, especially in high-resolution videos. The framework's real-time capabilities on different hardware configurations indicate its suitability for resource-constrained environments. The suggested method has great potential for surveillance, autonomous systems, and remote sensing, where real-time and precise object detection in high-resolution movies is crucial.