Semi-automatic Segmentation on superpixel representation of N-D Image

July 2016 - current

Image Segmentation: Classifying an image into meaningful objects such as background and foreground. This project is an attempt on image segmentation using Boykov Kolgomorov graph partitioning algorithm and CIEDE2000 color difference formula. Superpixel representation of the image is obtained using (Simple Linear Iterative Clustering). Core features of OpenCV library is utilized for the implementation.

DualIso

An Algorithm for Subgraph Pattern Matching on Very Large Labeled Graphs. Implemented a method which uses fine pruning techniques to make sub-graph isomorphism possible, regardless of its NP-complete nature. Implementation was done as part of course work during post-graduation in NITK.

Interactive Robot: Prototype

Designed and developed an Interactive Device using ATmega328 microcontroller with help of open-source hardware project Arduino. Different sensors were used to collect the environmental awareness. Collected data using sensor-fusion were used to take decisions. Different locomotive and signalling actuators were helping the device to interact with ourtside world. This work was Inspired by a TED Talk by Massimo Banzi (founder of Arduino Project). This talk intrigued us to think about the possibilities of Interactive Devices. The things around us, sharing common world amoung us, what if they could talk on how they feel about sorroundings. It's cool when a plant is intelligent enough to tweet about its well being (consider: I'm groot !).