EVL faculty and Computer Science assistant professor Wei Tang has been working with Motorola to use computer vision to improve cell phone photos taken in low-light conditions. Dr. Tang’s research focuses on the integration of deep neural networks with conventional image processing techniques to address complex computer vision problems. Specifically, Tang and his graduate student researchers are working on methods to deblur low-light cell phone images, to return a crisp, clean version of the picture. The process includes providing the neural network a series of images - both blurry and clean - to create a loss function, or learning objective, to identify the preferred images. The resulting deblurred images are then provided to the Motorola team to brighten the low-light photos for highly improved results.

Tang’s research interests in computer vision, pattern recognition and machine learning can also be applied to augmented reality and intelligent robotics applications. Stay tuned for updates in Tang's research and collaborations.

Read the related CS e-newsletter article by Andrea Poet for more detail.

URL