Masatoshi Okutomi

V-ECE allows clinicians to customize lesion visibility in endoscopic images without the need for time-consuming dye application, enhancing diagnostic efficiency.

CPDDNet achieves superior image quality and polarization accuracy by seamlessly integrating denoising and demosaicking, revealing the untapped potential of joint processing in color-polarization imaging.

The GastroNVS dataset could revolutionize gastroendoscopic imaging by enabling advanced view synthesis techniques that enhance both training and procedural outcomes.

Leveraging NeRF-rendered images, this method enhances 3D Gaussian Splatting to achieve high-quality street scene rendering without sacrificing speed.

Ditch 2D tracking limitations: this new approach uses 3D geometry to boost object tracking in street-level imagery by 22%.