Jianbo Shi

Jianbo Shi is a professor of computer and information science at the University of Pennsylvania whose normalized cuts method treats image segmentation as graph partitioning. Born in Shanghai, he earned a BA in computer science and mathematics at Cornell in 1994 and presented “Good Features to Track” with Carlo Tomasi at CVPR in June 1994. At Berkeley, he developed the normalized cut criterion with Jitendra Malik in 1997 and completed his PhD in 1998, and their normalized cut paper won the IEEE Longuet-Higgins Prize in 2007. After leading the HumanID project as research faculty at Carnegie Mellon's Robotics Institute from 1999, he joined Penn in January 2003. Later work includes the Ego4D and Ego-Exo4D egocentric video datasets and a Memory Encoding Model that won the Algonauts 2023 visual brain competition.

Now at: University of Pennsylvania
Works: 137
Badges: Loyal

Publication numbers from OpenAlex, as of 22 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

1994

Good features to track

Born in Shanghai, he earns a BA in computer science and mathematics at Cornell University in 1994. In June 1994, he and Carlo Tomasi present “Good Features to Track” at CVPR, proposing a feature selection criterion based on how the tracker works.
1997–1998

Normalized cuts at Berkeley

At UC Berkeley's EECS Department, he and Jitendra Malik write a May 1997 technical report that treats image segmentation as a graph partitioning problem and introduces the normalized cut criterion. He receives his PhD in computer science from Berkeley in 1998 for a thesis on the normalized cuts segmentation algorithm.
1999–2003

Research faculty at CMU

He joins the Robotics Institute at Carnegie Mellon University in 1999 as research faculty. There he leads the Human Identification at Distance (HumanID) project, developing vision techniques for identifying people and inferring their activity.
2003–2010

Joining Penn's CIS department

In January 2003, he joins the Department of Computer and Information Science at the University of Pennsylvania, where his GRASP Lab group builds algorithms to understand human behavior and interaction with objects in video. With Ben Taskar, he takes part in the 2010 ONR MURI “Rich Representations with Exposed Semantics for Deep Visual Reasonings,” led by CMU.
2005–2007

The Longuet-Higgins Prize

He receives an NSF CAREER Award in 2005 and the IEEE Longuet-Higgins Prize in 2007. The prize, a test-of-time award, goes to the normalized cut paper with Jitendra Malik, a general clustering algorithm that has inspired much work on graph partitioning.
2021–2023

Ego4D and Ego-Exo4D

In October 2021, he is among the coauthors of Ego4D, a massive-scale egocentric video dataset and benchmark suite of daily-life activity. Ego-Exo4D, which he also coauthors, follows in November 2023 with simultaneously captured egocentric and exocentric video of skilled activities such as sports, music, dance and bike repair.
2023–2024

Winning Algonauts 2023

With Huzheng Yang and James Gee, he builds the Memory Encoding Model, which adds memory-related information as input and wins the Algonauts 2023 visual brain competition. Their paper “Brain Decodes Deep Nets” is a CVPR 2024 highlight.
As of October 2026

Professor at Penn

He is a professor of computer and information science at the University of Pennsylvania and a GRASP Lab faculty member, working on computer vision, machine learning and dynamic intelligence. His open-source package ncut-pytorch runs Nyström normalized cuts on million-scale graphs in milliseconds.