Automated Marine Debris Detection: A Deep Learning and Ensemble Approach with Web Deployment

Objectives: To propose a scalable, real-time detection framework for existing ocean debris detection systems using deep learning. Method: Convolutional Neural Networks (CNNs) and Random Forest (RF) classifiers are combined to increase the detection accuracy and to provide a robust monitoring solution. The performance of the CNNs, the RF classifier, and their ensemble is compared to evaluate their performance. Findings: The results show that the ensemble model outperforms individual models by achieving a higher accuracy and F1 score of 95.1% and 0.95, respectively. Furthermore, object detection is performed using the Clarifai API, integrated into a Voila-based web application. Novelty: The web based application allows users to upload images and receive detection results. The proposed method provides a practical, scalable tool for monitoring and mitigating plastic pollution in the world’s oceans. Keywords: Convolutional Neural Networks, Random Forest classifier, Ensemble algorithm, Debris detection system, Clarifai API