Infrared Small Object Detection using Hybrid Augmented Intensity and Density based Clustering

Small object detection is the process of identifying tiny physical entity or features present within an image. Detecting small objects in infrared images presents unique challenges due to limited visible features and signal quality. The existing study introduces an established method for spotting small targets in single-frame IR images. The existing method detects the infrared images and the drawback is that the method only detects the images and does not label it. In order to overcome the above drawbacks, the proposed method concentrates on two key elements: improving infrared intensity maps using careful analysis of variation and applying clustering techniques to recognize object shapes and central points. These steps ensure precise detection, even for very small targets. The proposed integration of the YOLOv5 algorithm improves both speed and accuracy, enabling real-time detection in complex infrared scenes. The proposed method detects accurately up to 95%, which is higher when compared to the existing method along with the labeling of objects. This approach offers quick and precise small object detection in infrared images with potential benefits in surveillance.