Nishit Parekh

Vision-Projects

Work at Eyenuk and Perceptive Automata

My full-time work at these companies is not shown here, due to some ambiguous company policies. While I get that sorted, I am more than happy to talk to you about it over a call or in person. Please contact me to learn more!

Real-Time Scoreboard Detection and Recognition in Sporting Event Videos

Activity Recognition
Activity Recognition

Internship Project, Gracenote (Summer 2017)

I designed a system to extract team scores and game clock information from videos of sporting events in real-time. The system was built to be robust to occlusions, ad-breaks, and sudden changes in game scenarios. The detector worked on a variety of sporting event videos (Basketball, Hockey, Baseball etc.) on a range of sources and their custom scoreboards (ESPN, YouTube, Fox Sports etc.).


Activity Recognition in Sport Videos

Activity Recognition Activity Recognition

Course Project, Computer Vision (Fall 2016)

We built an activity recognition system to classify actions performed by Volleyball players during a match, by leveraging views from an on-court camera and available training data for 6 classes of actions (“serve”, “smash”, “block” etc). We used two approaches for the problem: One using hand-crafted Optical Flow and HoG features classified using an SVM, and another approach using a CNN on raw frames.


TMMS in Colour Images

Using colour-based TMMS to improve Text Segmentation Using colour-based TMMS to improve Text Segmentation

Research Project, LRDE, EPITA (Summer 2015)

I worked on improving the TMMS morphological operator (used for text segmentation) by devising a partial-ordering for colours based on the background of the text, to utilise the same TMMS algorithm but in a colour domain, thereby improving the text segmentation results. The results were based on the ICDAR Robust Reading Challenge Natural Scenes Dataset.