Monitoring and analyzing object activities in video scenes is a cornerstone of surveillance, providing actionable insights to meet specific objectives. In underwater environments, this task becomes particularly challenging due to the need for precise detection of moving objects, compounded by the complex visual distortions inherent to underwater settings. Existing methods often fall short in preserving critical attributes, leading to suboptimal detection performance. Inspired by the remarkable learning capabilities of neuromodulators like dopamine and norepinephrine in the human brain, particularly in the detection of objects like tasks, we propose an innovative adversarial learning-based architecture for object detection in underwater video frames. Our approach outperforms eight existing methods on the benchmark dataset, namely Underwater change detection, demonstrating superior performance in both qualitative and quantitative evaluations. The proposed method achieves overall precision, recall, and f-measure of 0.98, 0.97, and 0.97, respectively, demonstrating significant performance over state-of-the-art techniques.