Underwater Image Enhancement Using CycleGAN with Attention Mechanism

In this research work, we study the problem of poor quality underwater images caused by light absorption and scattering in water.Underwater images usually appear dark, unclear, and color distorted, which affects underwater observation and analysis.To solve this problem, we propose an underwater image enhancement method using a modified Cycle GAN model with attention mechanisms.Our approach improves image clarity, contrast, and color balance by learning from unpaired underwater images.Attention mechanisms help the model focus on important parts of the image during enhancement.The system is trained using different underwater images so that it can work well under various water conditions.For practical use, we implement the proposed model using a Django backend and a Flutter mobile application, allowing users to upload or capture underwater images and obtain enhanced results.Experimental results evaluated using PSNR and SSIM show that the proposed method produces visually improved images and performs competitively when compared with existing enhancement techniques.