A Novel Approach on Automatic Maritime Object Detection and Classification Using Support Vector Machine in Comparison with Residual Networks

This research aims to evaluate the performance of these two approaches in a curated dataset that reflects the challenges of detecting and classifying maritime debris. By comparing SVMs and ResNets, this work seeks to provide insights into the strengths and limitations of each method, offering guidance for researchers and practitioners working towards more efficient and accurate maritime surveillance systems. Materials and Methods: In this work, we applied both the SVM algorithm and the ResNet algorithm, each involving 10 iterations (N=10). These two algorithms were assessed across two distinct groups, and a total of 100 samples were considered in the analysis. The Gpower statistical test was employed, with a power setting of 85% (g power parameters configured with
$\boldsymbol{\alpha}=\mathbf{0.05}$
and power=0.85). This choice of power setting was made to ensure that the study had a robust ability to detect statistically significant differences or effects, in line with established statistical standards. Results: The research outcomes reveal a notable accuracy of 93.21% for Support Vector Machines (SVMs) and 91.73% for Residual Networks (ResNets) in maritime object detection and classification. This underscores the effectiveness of SVMs, suggesting their practical applicability in real-world scenarios for tasks like environmental conservation and sustainable marine practices. Conclusion: In conclusion, our research lays the foundation for advancing maritime surveillance with SVMs and ResNets, acknowledging current limitations and emphasizing the need for ongoing improvements to foster sustainable marine practices and environmental conservation.