Sumesh Thakur

782-***-**** add8zy@r.postjobfree.com www.linkedin.com/in/sumeshthkr sumeshthakr.github.io Profile

2nd year master’s student at Saint Mary’s University, Halifax. Part of Graphics & Spatial Computing lab. Passionate about computer vision and deep learning. Highly capable leader, having led multiple group projects to completion. Proficient in a range of modern technologies including Python, C++, JAVA, and JavaScript. Expertise in modern deep learning libraries including PyTorch, Keras and TensorFlow. Education

MASTER OF SCIENCE IN MATHS & COMPUTER SCIENCE APRIL ‘19 - PRESENT SAINT MARY’S UNIVERSITY, HALIFAX, NOVA SCOTIA

· Working towards developing a robust and accurate object detection system for vehicles through 3D LiDAR point clouds using deep learning as part of the master’s thesis.

· Worked on various projects like objects segmentation and clustering of point clouds, 3D fully convolution layers for pedestrian detection in road scene, sparse convolutions for feature learning from Point clouds.

· GPA 3.7 / 4.3 (until May 2020)

· Supervisor: Dr. Jiju Poovancheri

BACHELOR OF TECHNOLOGY IN COMPUTER SCIENCE ENGINEERING JULY ‘14 - JULY ’18 CHANDIGARH ENGINEERING COLLEGE, I.K GUJRAL PUNJAB TECHNICAL UNIVERSITY, MOHALI, PUNJAB

· Lead Coordinator of Computer Science Engineering Gaming Club. o Lead a team of 6 coordinators for organizing inter university video game competition, 2017, with participation from 10+ universities with 600+ participants.

· GPA: 75% (with Distinction)

Technical Skills

· Broad experience in Modern Computer Vision, Deep Learning, Convolution Neural Networks Algorithms, Graph Representation Learning, Mobile app development, Machine Learning Algorithms and Web Technologies.

· Programming Languages: Python3, C++, JAVA, JavaScript, HTML

· Software Libraries and Frameworks:

o Deep Learning and Machine Learning: Pytorh, Keras, Tensorflow, Sklearn o Point Cloud Processing: PCL, Open3D

o Data Visualization: MatplotLib

o Computer Graphics: OpenGL

o Optimization and Calculations: Numpy, Numba, Pandas o Mobile and Web Development: Android Studio, VueJs Work Experience

INNOVATION ASSOCIATE BARRIERBREAK SOLUTIONS DEC 2017 – APRIL 2019

· Involved looking at how to make the organization better, by improving existing tools, processes, and technologies. Involves delving into aspects of Mobile Applications, deep learning, augmented reality, NLP, and more.

· Lead the development of "6by6" Android Toolkit, first Indian currency detection application using deep learning. Relevant Courses

· CSCI-6671: Computer Graphics, Saint Mary's University.

· CSCI-6691: Special Topics in Real-Time 3D Vision (Directed Studies), Saint Mary's University.

· BTCS-701: Artificial Intelligence, Punjab Technical University.

· BTCS-503: Design & Analysis of Algorithms, Punjab Technical University.

· CS224W: Machine Learning on Graphs (Stanford University), MOOCs.

· CS231N: Convolutional Neural Networks (Stanford University), MOOCs Scholarships & Rewards

· FGSR Fellowship, Saint Mary’s University, 2019.

o valued at $ 17,500 per year

· First Position, 3 Minute Thesis Competition at Saint Mary's University, 2020. o Won $750 as winning prize, competing with 19 participants from master’s and PhD program. o Representing Saint Mary’s University in Eastern Regionals (Postponed due to COVID-19).

· Instructional Skills Workshop Certified Instructor, 2019. o Completed certification for ISW instructor program, a comprehensive three-tiered instructor development program that serves as the foundation for several professional development activities. Projects

· Professional Projects

o “6 by 6” Android Application

Developed three In-App Modules namely Currency Detector, Light Detector and Magnifier for visually impaired.

Created “Indian Currency” dataset from scratch, consisting “10 currency classes” and 100000 images.

Employed Inception Net (MobileNet) architecture for currency classification, on Android platform.

Currently application has more than 5000+ downloads with an average 4.0+ rating.

· Academic Projects

o Bird Detection in Wild for Shape Matching

Developed a bird detection tool in wild for analyze and study bird formations.

Created a “bird” dataset consisting of more than 1200 bird images with more than 4500 manually annotated objects (birds, airplanes, kites etc.)

Employed SSD architecture at backend for object detection. o Cornell Box generation using Raytracing.

Constructed Cornell box as part of CSCI-6671 course, using raytracing with features like Reflections, Diffused Shadows and Antialiasing.

o Procedural Landscape creation using Perlin Noise.

Constructed a virtual world landscape using Perlin noise in OpenGL. o Poisson Grid Sampling in 3D Pointclouds.

Modified Poisson grid sampling algorithms to 3D points. o 3D point segmentation and clustering using RANSAC and K-D Trees

Performed object clustering and segmentation on KITTI LiDAR dataset samples. Publications

Bivash Pandey, Sumesh Thakur, Hemanchal Joshi, Ayusha Pradhanga, Yasushi Akiyama and Jiju Peethambaran” Towards Video based Collective Motion Analysis through Shape Tracking and Matching” (Under Review)

Contact this candidate