I joined the Computer Science at the University of North Carolina at Greensboro in 2025. I completed my Ph.D. in Computer Science and Software Engineering at Auburn University, where I was advised by Dr. Sathyanarayan Aakur. My research focuses on computer vision and artificial intelligence, with an emphasis on graph-based representations for visual understanding. During my Ph.D., I explored how structured knowledge in the form of graphs can enhance a machine’s ability to perceive and reason about visual scenes. My work aims to bridge the gap between low-level visual data and high-level semantic understanding. I am passionate about teaching, mentoring, and building intelligent systems that are both explainable and adaptable to real-world challenges.

Bio-medical Object Detection (2020):

Developed a dataset and evaluation platform for object detection in canine cutaneous mast cell tumors, particularly in low-resource settings.

Open-World Action (2021):

Explored open-world action recognition in egocentric videos using an energy-based pattern theory framework.

Scene Understanding (2022):

Generated spatial graphs of static scenes using a graph generative approach by first generating the graph structure and then labeling the edges.

Open-World Action with Visual Grounding (2023):

Used vision-language models and commonsense reasoning to ground objects in egocentric videos and inferred actions using an energy-based pattern theory framework.

View-Invariant Egocentric Representation (2024):

Ongoing project focused on learning view-invariant representations of egocentric videos for understanding long procedural tasks.

• EASE: Embodied Active Event Perception via Self-Supervised Free Energy Minimization.
  Zhou Chen, Sanjoy Kundu, Harsimran S. Baweja, Sathyanarayanan N. Aakur – IEEE Robotics and Automation Letters, 2025.
• CALIBFREE: Self-Supervised Feature Disentanglement for Calibration-Free Multi-Camera Multi-Object Tracking.
  Ruiqi Xian, Iain Melvin, Deep Anil Patel, Sanjoy Kundu, Martin Renqiang Min, Dinesh Manocha – In Review, 2025.
• ProbRes: Probabilistic Jump Diffusion for Open-World Egocentric Activity Recognition.
  Sanjoy Kundu, Shanmukha Vellamcheti, Sathyanarayan N. Aakur – Accepted at ICCV 2025.
• Hallucinate, Ground, Repeat: A Framework for Generalized Visual Relationship Detection.
  Shanmukha Vellamcheti, Sanjoy Kundu, Sathyanarayan N. Aakur – In Review, 2025.
• Discovering Novel Actions from Open World Egocentric Videos with Object-Grounded Visual Commonsense Reasoning.
  Sanjoy Kundu, Shubham Trehan, Sathyanarayanan N. Aakur – European Conference on Computer Vision (ECCV), 2024.
• IS-GGT: Iterative Scene Graph Generation With Generative Transformers.
  Sanjoy Kundu, Sathyanarayanan N. Aakur – CVPR, pp. 6292–6301, 2023.
• Knowledge Guided Learning: Open World Egocentric Action Recognition with Zero Supervision.
  Sathyanarayanan N. Aakur, Sanjoy Kundu, Nihil Gunti – Pattern Recognition Letters, 2022.
• Benchmark and Evaluation of Low Resource Object Detection in Biomedical Images.
  Sanjoy Kundu, N. Gunti, B. Hendrickson, S. More, S.N. Aakur – IEEE AIPR, 2020.
• Gain and SAR Improvement of a Conventional Patch Antenna Using a Novel Pi-shaped DNG Metamaterial.
  Anik Mallik, Sanjoy Kundu, Md. Osman Goni – IEEE EICT, 2014.
• Design of a Novel Dual-Band Microstrip Patch Antenna Operating at 2.45 GHz and 2.84 GHz with Practical Implementation.
  Anik Mallik, Sanjoy Kundu – IEEE ICCIT, 2014.
• Design of a Novel Two-Rectangular U-shaped Double Negative Metamaterial.
  Anik Mallik, Sanjoy Kundu, Md. Osman Goni – IEEE ICIEV, 2013.
• An FPGA-Based Semi-Automated Traffic Control System Using Verilog HDL.
  Anik Mallik, Sanjoy Kundu, Md. Ashikur Rahman – ICECTE, 2012.

Download CV

Email:s_kundu3@uncg.edu