I am Abhinav Kulkarni. I have a keen interest in solving problems in the domain of embedded computing. Possess diversified educational background and specific industrial experience.

Concise summary of key research projects:

Research

Study on novel techniques to improve efficiency of signal detection in next-generation communication systems.

Location: LSSI Lab, UQTR, QC, Canada

  • With the surge in next-generation communication system applications, there is an increasing demand for advanced capabilities of computing circuits catering them.
  • The overall work addresses the challenge of developing integrated techniques for next-generation communication systems to achieve controlled trade-off between following KPIs :
    1. Energy consumption
    2. Computational complexity
    3. Interconnect Bandwidth
    4. Throughput
    5. Latency
    6. Signal Fidelity
  • Algorithm decentralization and approximate computing approaches are utilized in the work for SISO/MIMO signal detection.
  • Evaluation of techniques on FPGA computing platform.

Improving Accuracy of Superpixel Segmentation using Color Histogram-based Seed Initialization

Location: HESL Lab, NTU SCSE, Singapore

  • Superpixels are perceptually meaningful image segments formed by grouping pixels based on common properties, such as color, and are useful for accelerating computer vision applications.
  • In this work, the Histogram Super Linear Iterative Clustering (SLIC) algorithm is prototyped to enhance the accuracy of the SLIC algorithm by using a color histogram during seed initialization. The quantized color histogram captures localized color profiles within uniform grids in the raw image.
  • The quality of initial seeding can be controlled through user-defined parameters, such as histogram suppression window and maximum seeds per grid box.
  • Label enforcement of erroneous pixels is implemented on the superpixel segmented image by assigning a pixel with a particular cluster label based on the majority of surrounding labels.

GitHub

My Image

Braille Ebook Reading Device

Personal Learning Project

  • The novel prototype of the device models an ebook reading device for blind individuals using two tactile refreshable Braille displays.

Watch a demo video:

GitHub Paper