Source-Free Model Transferability Assessment for Smart Surveillance via Randomly Initialized Networks
A source-free unsupervised transferability assessment based on RINN and embedding similarity tailored for surveillance data to identify optimal general-purpose model for adaptation.
Jun 20, 2025
Event based surveillance video synopsis using trajectory kinematics descriptors
This paper proposes a novel event-based video synopsis method that addresses object occlusion by using trajectory kinematics descriptors and affinity propagation to cluster similar kinematic events, resulting in clearer and more efficient summary videos compared to state-of-the-art techniques.
May 8, 2017
Bridging the gap between lab research and market value.
I led the commercialization of the lab’s core research by developing a market-ready prototype and a viable business model. My role involved translating technical algorithms into a user-centric solution for surveillance efficiency. This venture won the APICTA Merit Award (R&D category) and secured funding, leading to a successful acquisition of the product/team by our industry partner, EverSTek. This experience honed my ability to identify user needs, build MVPs, and execute a successful technology exit. (Details to be followed)
Jun 30, 2016
Trajectory kinematics descriptor for trajectory clustering in surveillance videos
To overcome the problem of comparing object trajectories with varying lengths, this paper proposes a novel trajectory kinematics descriptor using Frenet-Serret frames, which achieves superior F-measure scores in event clustering compared to existing state-of-the-art methods.
May 24, 2015