Danny Tran

I am an incoming PhD student at UT Austin. Prior to this, I was an undergraduate Computer Science student at UC Berkeley where I had the pleasure of being advised by Professor Trevor Darrell and Dr. Amir Bar.

I'm interested in computer vision, robotics, and deep learning. I am primarily interested in enabling high level planning for robotics. Some of my work is highlighted below.

Email  /  CV  /  Scholar  /  Twitter  /  Github

Research

We present Navigation World Models. We plan trajectories by using a video world model, enabling planning, conterfactual reasoning, and action.

We present EgoPet, a new egocentric video dataset of animals. We create a variety of tasks to learn from animals such as Visual Interaction Prediction, Locomotion Prediction, and Vision to Proprioception.

Class Projects

CS280 is the graduate computer vision course at UC Berkeley. For my CS280 final project, I worked on a text-conditioned extension to the Locomotion Prediction task in my EgoPet paper.

CS285 is the graduate Deep Reinforcement Learning course at UC Berkeley. For my CS285 final project, I worked on implementing various deep-reinforcement learning algorithms for learning a bus policy to avoid bus bunching. I explored the impact of Domain Randomization on training such a policy.

CS182 is the Deep Learning course at UC Berkeley. Designed a deep neural network for classifying gender from CT scans, achieving an accuracy of approximately 80%. Implemented and analyzed Grad-Cam visualizations in order to determine facial features.

Miscellanea

As an academic intern for CS61B, the Data Structures and Algorithms course at UC Berkeley, I aided students during lab and office hours on various projects and homework. Created lab slides and code demonstrations which I presented in lab.

Website template taken from Jon Barron at source code. Thank you!