Engineer and wannabe economist with 9+ Years of experience in the field of Computer Vision, Machine Learning and Deep Learning. Graduated with Bachelors in Electrical and Electronics Engineering and Masters of Science in Economics from Birla Institute of Technology and Science, Pilani Rajasthan. Currently working in Amazon as a Applied Scientist II in the field of computer vision and machine learning. I like the challenge of developing complex vision/ Machine Learning solutions for memory/resource constrained hardware. Earlier in a different life, I worked in Canon on Vision framework which is used in many product families. Before Canon I have successful stint in Samsung. Successfully commercialized computer vision/ imaging solutions to Samsung Galaxy series mobile devices. We also developed POC of pixel wise image classification (semantic segmentation), virtual human character modelling and face feature tracking algorithms.
My github profile contains the following personal projects I worked on.
- Parametric Human Body Modeling
- Image recognition server using nodejs and OpenCV.
I contributed to STASM face feature tracking open source project
- http://www.milbo.users.sonic.net/stasm/
I am interested in computer vision and machine learning and have 3 patents and 2 papers in that domain. I am excited about the possibilities of applying machine learning/ deep learning algorithms to “big data”. Details are given below.
Patents and Publications
- Method and apparatus for coding unit partition search (Indian patent filed). The patent deals with a new algorithm for detecting coding unit partition in a fast and efficient way.
- Composition based Gallery Image search using improved semantic Segmentation US9886762B2. The patent deals with a new way of searching local/cloud images using semantic information.
- Virtual Reality Headset controlled by Smart Watch US10521009B2. The patent deals with the use of rotating bezel of a smartwatch to control the Virtual Reality Headset.
- Fast Coding Unit Partition Search. The paper presents a novel algorithm to find a good coding unit partition which (CU partition) results in good compression similar to brute force search. Paper published in IEEE ISSPIT Vietnam 2012.
- Single Touch Segmentation using Random Walks and Graph Cuts. Paper published in International Conference on Image Processing (ICIP 2015).
- Generic Approach for Document Extraction using Semantic Segmentation Badrinath G. Srinivas, Satish Lokkoju and Amaresh K Pande. Paper published in ACVC 2020.