Underwater mines are common explosive bombs that are placed around a coastal region of a country, deep under the ocean to ensure safety, defense and security. This could be a threat to a submarine or any civilian boats that crosses over the mine. Though it provides enormous security, it is also a threat to marines which may mistake it as a rock. Due to its similar appearance, size, and placings in ocean bed, a rock can appear to be and mine and vice versa. Moreover, absence of light under ocean due to attenuation makes it impossible to capture an image to be processed. Here, we use the sonar technique to explore the ocean bed and identify rocks and mines accurately. The dataset contains 60 different angles under which rocks and mines are recorded with specific value. The main objective is to enhance the model using Ensemble learning techniques. Standalone algorithms to overcome overfitting, under fitting and feature selection. Bagging to improve stability and reduce bias value.