Efficient marine debris infrastructures on optimising SVM with LoG segmentation for enhanced IoR, DC and Hausdorff distance performances

In the face of escalating threats to aquatic ecosystems posed by marine debris, the demand for precise and efficient classification techniques becomes paramount. This study employs image segmentation methods Canny edge detection, Sobel operator, and Laplacian of Gaussian (LoG) to partition photographs of maritime trash. A notable addition is the integration of SVM-based classification, offering promising avenues for environmental surveillance and disaster management. By incorporating the LoG process, the identification of blob-like structures enhances the accuracy of debris segmentation. Comparative analysis utilising metrics like intersection over union (IoU), dice coefficient, and Hausdorff distance underscores the efficacy of the combined LoG and SVM approach. This synergistic method adeptly detects edges via the LoG operator and ensures accurate debris classification through SVM modelling. The results demonstrate significant improvements, yielding higher IoU (0.993), dice coefficient (0.996), and minimal Hausdorff distance (0.0000977). Executed in Python, this research propels marine debris analysis forward by furnishing a robust framework for automatic image categorisation, which is vital for initiatives aimed at environmental preservation.