Object identification in SONAR images is crucial for underwater exploration, submarine rescue operations, hostile object reconnaissance, etc. SONAR systems are reliable imaging sources in environments where optical methods fail due to low visibility, turbidity, or complete darkness. An accurate and efficient model to detect, classify, and track the objects in SONAR imagery may ensure the accomplishment of these underwater tasks. Object detection in SONAR images is difficult because the images are noisy and have low resolution. Although Convolutional Neural Networks (CNNs) are used for this purpose, these approaches face some challenges due to the lack of large labeled datasets. This manuscript focuses on two major tasks: first, object detection, and second, classification. The YOLOv8 model has been utilized for object detection, and further, a parallel CNN has been integrated with it for object classification. For comparison of results, various deep learning models have been used, and the results are analyzed on two well-known datasets in terms of classification accuracy.