| Google Scholar | Github | |
I am a computer vision researcher at Lawrence Livermore National Lab (LLNL), focusing on object localization and instance retrieval for applications including GeoAI, person re-identification, and materials recognition. I completed my PhD in EECS (Video and Image Processing (VIP) Lab) at University of California, Berkeley, under Prof. Avideh Zakhor. My dissertation was titled, "Representation Learning for Efficient Localized Image Retrieval", which will be available on this page when the embargo is lifted. |
|
|
We describe and demonstrate the Gallery Filter Network (GFN), a novel module for person search, which can efficiently discard gallery scenes from the search process, and benefit scoring for persons detected in remaining scenes. |
|
We show sensors designed for human recognition do not produce imagery optimal for deep learning algorithms, and develop a framework for optimizing for automated recognition. |
|
Class project studying gradient descent on the triplet margin loss function, a non-convex objective function used for metric learning. |
|
Class project exploring rapid prototyping of hand action recognition systems for RGB + depth video data from the CamBoard pico monstar camera. |
|
Class project investigating whether image super-resolution can be used to enhance the discriminative features of imagery, such that the transformed imagery is more amenable to classification. |
|
Project at LLNL exploring fine-grained recognition of vehicles in overhead imagery using deep learning models and the triplet margin loss function. |
|
A deep learning tutorial I gave with Cindy Gonzales as part of the Winter Hackathon at LLNL in February 2021. |
|
An introductory deep learning class I helped teach at LLNL in January 2019. |
Website design: ✩ |