Computer Vision Lab

The Rutgers ECE Vision Lab led by Dr. Kristin J. Dana conducts research in computer vision at the intersection of machine learning and computational photography. In our research, the term “camera” is used in a very broad sense to mean the collection of optical, mechanical and electrical components to send, sense and interpret light. Computational appearance capture includes the paradigm where machine learning affects appearance capture, optimizing for the task at hand. In my current work, example cameras include: systems incorporating robots for motion control; cameras with additional optical elements with unusual shapes to bend light advantageously; and camera-illumination systems that use illumination to communicate messages to the camera element. This work naturally lends itself to interdisciplinary science that crosses field boundaries such as communications, robotics, civil engineering, and medicine.

2019 CVPR paper Light Field Messaging with Deep Photographic Steganography

Two CVPR papers have been accepted for 2018, Context Encoding for Semantic Segmentation(Oral) and Deep Texture Manifold for Ground Terrain Recognition
We have released the code for Deep Encoding
We have released the GTOS database Ground Terrain in Outdoor Scenes

Two CVPR papers have been accepted for 2017, Deep-Ten Texture Encoding Network and Differential Angular Imaging for Material Recognition

Copyright © Computer Vision Lab, Rutgers University.