Coastal Marine Debris Identifier

There is an ever-increasing interaction between man and the marine environment, which makes the problem of marine debris more significant. This study focuses on devising a system for locating such debris. Use knowledge transfer beyond single task transfer learning to leverage the potential of different models and utilize available large datasets to improve detection efficiency. Construct and refine a deep learning system based on convo neural network paradigms to solve the problem of marine debris detection by employing very high-resolution space images. Satellite imageries are fetched from MAXAR Satellite images. The debris differentiates depending on the depth of the object, density of objects in the debris, and degradation of the structure. This paper describes the new, highly practical model, especially in the field of coastal debris collecting from satellite images; based on the Convolutional Neural Network methods with incorporated Deep Transfer Learning and model finetuning. Rationally, it develops, extends, and tests a multimedia system focusing on the fine detailing of the detection of marine debris that assures performance by standards set within the scope of CNNs. Deep Transfer Learning applies a fine-tuning approach to a pulled network (trained on too many photos of non-related objects), speeding up performance and leading to higher marine debris detections. Thus, all stress the revolutionary nature of using CNN techniques within the framework of deep learning networks for the improvement of automatic systems that aim at the detection of coastal marine debris.