Abstract

Developing an effective underwater object detection method is crucial for both the conservation and utilization of marine life. However, the complexity and unpredictability of underwater environments introduce numerous challenges into the field of underwater object detection. The exploration and monitoring of underwater environments present unique challenges due to factors such as poor visibility, Color distortion, and varying lighting conditions. There have been numerous automatic techniques devised to tackle these challenges. Deep learning algorithms like Faster R-CNN, YOLO, SSD and machine learning algorithms which include SVM, RF, K-NN, Naive Bayes etc., which can be implemented over a large number of underwater images to detect objects or marine organisms present within them. For preprocessing the images, several Image processing techniques have been developed like Color correction, Contrast enhancement, Noise reduction etc. to optimize the detection process. This research paper discusses many object detection approaches in underwater images that have been proposed by researchers in recent years and challenges that are being encountered. In this study, we examined the performance of various deep learning algorithms used by various researchers in their paper in the context of underwater object detection