Advancing Marine Plastic Detection with CLAHE-Enhanced Underwater Images Using YOLOv8 Integrated Attention Mechanism

The accumulation of plastic garbage in aquatic ecosystems puts at risk aquatic habitats as well as the larger global ecosystem health, posing serious environmental and financial dangers. By introducing hazardous chemicals and upsetting fragile marine ecosystems, this pollution has an impact on human health and marine life via the food chain. The current approaches to identifying marine plastic are labor-intensive and ineffective in the face of the worldwide problem of aquatic garbage. The custom-trained YOLOv8 deep learning model is presented in this work, increasing the identification of marine and undersea trash. Furthermore, an Efficient Channel Attention (ECA) module is integrated with YOLOv8. Our proposed YOLOv8 model outperforms YOLOv7 and YOLOv6 in accuracy, precision, recall, and F1-score, even in the face of difficulties such as decreased visibility and warped object shapes. By tackling the serious environmental and financial issues raised by plastic pollution, deep learning demonstrates notable advancements in real-time detection. This research represents a significant advancement in automated, accurate, environmentally dangerous marine plastic identification. The results indicate that our proposed attention mechanism improved YOLOv8 model performed better than remaining state-of-the-art models with an F1-score of 97.8%, a precision of 98.12%, and an mAP score of 95.4%, highlighting its robustness in reliably identifying particular targets across a range of assessment metrics.