Abhishek Kumar

Computer Vision & Machine Learning engineer

Tessellate Imaging

Matlab 3 years
Python 3 years


I have always been an admirer of automation and intelligent systems. The need for these systems has increased primarily because we humans have found solace and peace in our own comfort zones. This attitude in-turn has completely destroyed our education system. We are taught to mug up the concepts, pour that batter in the exams to score high grades and forget everything we tried to learn. I have the vision to change these inefficient ways and to traverse towards that vision I am inquisitively researching in the field of machine vision and cognitive intelligence.
In my current venture, Tessellate Imaging Pvt Ltd, we are working on providing deep learning and machine learning based consultancy to companies. We build custom software services by addressing clients' concerns involving ML and DL in the field of computer vision.
My Publications -
1. Analysis and optimization of parameters used in training a cascade classifier
Published Journal, Advances in Image and Video Processing, Society for Science and Education, United Kingdom, ISSN: 2054-7412, Vol:3, Issue: 3, April 2015
Short Description:
– The paper is published in association with Ayonix - Face recognition company, Japan. Research focused on optimizing the parameters used in Local Binary pattern and Histogram of gradients based cascade classifiers.

2. A Novel System to Monitor Illegal Sand Mining Using Contour Mapping and Color based Image Segmentation
Published Journal, International Journal of Image Processing, Computer Science Journals, Kuala Lumpur, Malaysia.ISSN: 1985-2304, Vol: 9, Issue: 3, June 2015
Short Description:
– The work is done under the guidance of Whishworks, Hyderabad. The research was done on implementing a sand mining monitoring system
• Traffic Sign Recognition using Weighted Multi-Convolutional Neural Network

3. Under Review at IET Intelligent Transport Systems Journal

SOFT SKILLS:

Positive Attitude
  4.4 / 5
Team work
  4.5 / 5
Responsibility
  4.5 / 5
Flexibility
  4.6 / 5
Problem Solving
  4.6 / 5
Leadership
  4.4 / 5

WORK PERSONA

English communication
  4.6 / 5
Past work clarity
  4.7 / 5
Client interaction experience
  4.4 / 5
Transparency
  4.6 / 5
Open to learning
  4.7 / 5
Open source contribution
  4.2 / 5

INDUSTRIES SERVED

BusinessEducationNavigationPhoto & VideoTravel

PAST WORK

  • A Novel System to Monitor Illegal Sand Mining using Contour Mapping and Color based Image Segmentation

    Computer Vision Engineer

    Python Java OpenCV Tensorflow Xamarin

    This system includes a novel vehicle detection approach for detecting vehicles from static images and calculating the amount of sand being carried to prevent the malpractices of sand smuggling. Different from traditional methods, which use machine learning to detect vehicles, this method introduces a new contour mapping model to find important " vehicle edges " for identifying vehicles The sand detection algorithm uses color based segmentation since sand can have various colors under different weather and lighting conditions The proposed new color segmentation model has excellent capabilities to identify sand pixels from background, even though the pixels are lighted under varying illuminations

  • Face detection and Disguised Face recognition system

    Engineer

    Python Lua Torch PyTorch Caffe2

    A Deep Learning based implementation to for detection and counting faces in heavily occluded scenes and to distinguish disguised and imposter face

  • Real-time Basketball Statistics generation using Monocular Camera

    ML Engineer

    Python 3.6C++ Embedded C Tensorflow OpenCV

    Using a monocular camera setup, tracking basketball players throughout a game or practice session to generate player statistics in real-time.

  • Semantic segmentation and classification of Glioma cancer from histopathology slides

    ML Engineer

    Python Caffe DIGITS Theano OpenCV Scikit-Image

    Automating detection of cancerous regions of glioma cancer(Brain cancer) using Deep Learning based semantic classification of histopathology slides achieving state of the art results.