Abstract:Monitoring of pollution in water through connected smart devices is impracticable, still being an unexplored and far-fetched domain to investigate. Qualitative identification is typically used to depict the visible or aesthetic characteristics of water. Although the underwater environment remains unexplored and cumbersome to investigate, due to divergent prodigious activities, an underwater fleck detection method is proposed espousing the common visual features of water to investigate the underwater particulate matter. A 3D model for the proposed navigation unit and an application is modelled to determine the concentration of the particulate matter in the hydro samples collected from the surface, the mid, bottom level of a stagnant water body. The ML model is trained, and the execution time, RAM utilization, memory utilization, overall accuracy, model accuracy, and loss are analysed for the proposed system.

Underwater Fleck Detection Using Convolutional Neural Network