Underwater waste detection via quadruple neutrosophic image enhancement for marine environmental monitoring.

The increasing accumulation of waste in marine environments poses a serious threat to aquatic ecosystems and marine biodiversity. Effective detection and monitoring of this waste are crucial for environmental protection. However, underwater conditions such as poor lighting, turbidity, and color distortion make image-based detection highly challenging. Traditional enhancement methods often struggle to handle the visual inconsistency and unclear information present in underwater imagery, leading to reduced detection accuracy. To overcome these challenges, this paper introduces a novel Quadruple Neutrosophic Image Enhancement (QNIE) method aimed at improving both the visibility and accuracy of underwater waste detection. The proposed approach introduces four key components-truth, falsity, ignorance, and contradiction which enable effective handling of incomplete and conflicting visual information while enhancing critical image features such as edges and textures. Furthermore, when integrated with the YOLOv11 model, the proposed QNIE-YOLOv11 framework achieves 87% precision, 79% recall, and 84% mAP@0.5 on Underwater-Garbage-Debris Dataset, whereas the baseline YOLOv11 without QNIE attains only 79% precision, 64% recall, and 72% mAP@0.5. These results confirm that the proposed method delivers significant improvements in detection accuracy and visual quality over existing techniques, strengthening the reliability of automatic waste monitoring systems. This indicates that the proposed method not only advances underwater image analysis but also contributes to marine conservation and ecosystem protection, supporting broader environmental initiatives such as the United Nations Sustainable Development Goal 14 stating Life Below Water. All experiments were conducted using a fixed train-test split of the dataset, with images resized to a uniform resolution, standard data augmentation, and identical YOLOv11 training and inference settings to ensure fair and reproducible evaluation.