Abhishek Balasubramaniam

Hi, I am Abhishek Balasubramaniam

Graduate Research Assistant at Colorado State University

I am currently working on object detection model optimization techniques for low power perception with deep learning for emerging autonomous vehicles using Keras and TensorFlow. Experimenting with optimization techniques such as pruning, quantization, knowledge distillation, encoding and decoding, and co-optimization techniques for reducing the model size and memory allocation. We use Nvidia Jetson TX2 for testing and evaluation

 Area Of Research : 

Deep Learning
Model optimization
Connected and automated vehicles.
Reinforcement Learning
Image processing

 Highlights : 

Graduate Research Assistant:

I am working on deep learning model optimization techniques for low power perception for autonomous vehicles at the EPiC lab at Colorado state university under the guidance of Dr. Sudeep Pasricha. As a part of my research, I have been the Technical Advisor of the Connected and Autonomous vehicles system for the CSU vehicle innovation team for ECOCAR challenge.

Graduate Teaching Assistant:

I am the GTA for ECE 452 Computer Architecture and Organization taught by Dr. Sudeep Pasricha. I help in preparing course material, Assignments, resolving doubts and queries of students, and grading the course. The course consists of a total of 54 students.

Object Detection Model Compression for Resource constraint platform:

Recreated the SSD object detection model on using TensorFlow and Keras (not using TF object detection API). Worked on mixed-precision pruning, iterative layer by layer pruning and quantization. I trained this model as a graph using custom loss function and gradient function. Using train as a function.

ROS Based Stereo Vision System for Autonomous Navigation:

A Stereo vision based autonomous navigation system that used Deep learning model for Object Detection and Navigation. ROS For communication and an android app with ROS backend for GPS data acquisition.

CAVS Technical Advisor:

As a part of CSU vehicle innovation team for Ecocar Mobility challenge is a competition in which we develop a prototype of Level 2 automotive vehicle. My role as a graduate student is to come up with ideas that can improve the CAVS system performance.