If one knows what something is called, a text-based search on websites such as Google and Wikipedia are tremendously useful. However, what if one only has an image or mental image of that "thing" - how can one find it? Using text-driven searches for identifying images has so far proven very difficult. The ultimate goal for Dr. Serve Belongie of Cornell University is to fill this gap in online-search engines and revolutionize the way we search for and identify specific images. Visipedia, an innovative combination of computer vision, machine learning and crowdsourcing, bridges this gap. His research uncovers novel ways of capturing and sharing visual knowledge. By applying the tools of computer vision and machine learning, he will help people visualize the world around them in ways once thought unimaginable

• Dr. Belongie's research in computer vision technology draws upon the complementary strengths of humans and machines to solve an array of human needs.

• More specifically, he trains computers to serve as human visual-assistants for a wide range of applications. For example, he started the "Grocery Shopping Assistant for the Blind" initiative which provides assistive technology for the visually impaired. This technology grants access to textual information located in grocery store settings - thus, comparing two identical cans of soup is no longer a privilege of the sighted!

• Additional applications of Dr. Belongie's research include an electronic-eye to monitor sensitive coral reefs, video fire-detection systems, and facial recognition programs. At this time, the "Visipedia" project (in collaboration with Caltech) remains the top focus for his personal research.

• Today, Visipedia enables bird watchers to identify hundreds of species of North American birds. Once a photo of a bird has been submitted to the Visipedia application, a few questions regarding the bird's physical characteristics are proposed to the application user. Once the user provides answers, Visipedia can uncover the bird species and provide background information.

Ultimately, the Visipedia project aims to augment Wikipedia with search-by-image functionality for visual categories including, but not limited to, species of animals and plants, styles of clothing, and architecture. The goal is to enhance our human experience by teaching computers to see. 

Dr. Belongie's background in start-up companies, Digital Persona (a world leader in fingerprint biometrics) and Anchovi Labs (a photo annotation and management company that was recently acquired by Dropbox) helped him understand how to concentrate his research in ways that are commercially viable; however, he also enjoys discovering innovation that benefits education. His research for the Visipedia project has the potential to improve the functionality of a highly used research database (Wikipedia) not only for students, but for millions of people around the world.