Computer Vision

Computer vision research at Oregon State University is defined by its focus on embodied intelligence, treating visual perception as the foundational sense for machines that interact with the physical world. Our computer vision research rests on understanding objects and activities in 2D and 3D scenes (both still images and videos). We also work on tasks such as language-based scene navigation, video activity analysis, point cloud deep learning, video object segmentation, multi-target tracking, and explaining deep vision models so humans can understand how they make decisions.

Rather than analyzing images in isolation, the program integrates vision with robotics, natural language processing, and decision-making to create systems that can see, reason, act, and imagine the future of real-world scenes. Our computer vision research is also utilized in applications to analyze visual data in agriculture, forestry, molecular biology, and underwater sensing.