Specular reflections caused by bright light on wet surfaces in endoscopic images introduce significant visual distortions. These distortions obscure critical details, making automated image analysis and interpretation challenging. While existing methods employ traditional inpainting techniques to address this issue, the potential of state-of-the-art generative AI models, such as Generative Adversarial Networks (GANs), diffusion models, and transformers, remains largely unexplored. Our research aims to evaluate and compare various generative AI techniques for specular reflection inpainting in endoscopic images.
Click on the image below to watch our pitch video!
Click on our profile pictures to visit our LinkedIn profiles.