Pedestrian Behavior Prediction
A computer vision module with capability to predict whther a pedestrian is going to cross in front of an autonomous vehicle by estimating human intent using psychopysics and machine learning techniques.
About me
I am a Robotics grad student at University of Maryland, College Park. My career has revolved around developing software pipelines that solve real world problems using computer vision and machine learning techniques. My previous roles have been driven by motivation to solve one of the most difficult engineering problems, autonomous driving.
Tech Stack
Recent Work
Multi Task Learning with Vision Transformer
Implementation of original Segformer architecture that uses attention mechanism to perform image segmentation and augmented it to predict monocular depth using custom loss and upscaling
Human Detection Tracker
Use the Software Development Life Cycle process to design a computer vision system to detect and track humans in a video stream. The focus is on utiziling modern software practices of design and development.
Building built in minutes (3D reconstruction)
Developed a computer vision pipeline to reconsutruct a 3D scene using Structure from Motion techniques. Contributed specifically to thebundle adjustment and non linear triangulation modules of the pipeline
Get In Touch
Let's make something happen, contact me now!