UNDERWATER MINES DETECTION AND COMPARISON ON MACHINE LEARNING RESULT

Signals of sonars are progressively utilized in submerged quest and salvage for suffocating casualties, wrecks and planes. Programmed object order or recognition strategies can help a ton in the event of long hunts, where sonar administrators might feel depleted and subsequently miss the conceivable article. Be that as it may, the greater part of the current submerged article location strategies for sidescan sonar pictures are pointed toward recognizing mine-like items, overlooking the characterization of non military personnel objects, mostly because of absence of dataset. Also, considering the genuine dataset is imbalanced, we proposed a system to classification of object underwater whether object is mines or normal object . For classification there is served machine learning techniques is available. We will training module module based on 4 algorithms Logistics regression, ANN, SVM, Random forest . In this project we will classify objects and also compare accuracy results of these four, machine learning techniques.