Hasib Zunair

I am working on deep learning and its application in computer vision with Dr. Nabeel Mohammed from North South University. Previously, I was at The Tech Academy, where I worked on embedded input devices, robotics and computer vision.

I did my bachelors in Electrical and Electronic Engineering at North South University. My final year senior capstone involved an automated entry logging system using modern techniques, one of which involved real time facial recognition, where I was advised by Dr. Mohammad Monirujjaman Khan. During my undergraduate study, I was also involved with IEEE, where I served as the founding chair of IEEE Robotics and Automation Society Student Branch Chapter advised by Dr. Lamia Iftekhar.

Email  /  LinkedIn  /  GitHub  /  Google Scholar  /  Research Gate  /  Kaggle /  Hackerrank

My research interest lies in computer vision with a focus in deep learning. During my undegraduate study, I worked on mobile robotics, internet of things and embedded systems(motors and actuation) here. Below portrayed are my research work and projects; I've done and have been involved so far. Representative papers are highlighted. In my free time, I watch movies or play Counter Strike, sometimes also participate in Kaggle competitions.

Unconventional Wisdom: A New Transfer Learning Approach Applied to Bengali Numeral Classification

Authors : Hasib Zunair, Nabeel Mohammad, Sifat Momen

An accuracy of 97.09% on the NumtaDB Bengali handwritten digit datasets test set, which was obtained by freezing intermediate layers.

The unconventional approach used in this paper produces better results than conventional transfer learning while taking less epochs and having almost half the number of trainable parameters. Resulting in a more optimized model.

Estimating Severity from CT Scans of Tuberculosis Patients using 3D Convolutional Nets

Authors: Hasib Zunair, Aimon Rahman, Nabeel Mohammed

(10th place overall)

A 16-layer 3D convolutional neural network with a slice selection technique employed in the task of 3D Computed Tomography Image data of Tuberculosis(TB) patients which attained 10th place among a total of 100 participants in the ImageCLEF 2019 Tuberculosis - Severity scoring challenge. The goal is aimed at estimating TB severity based on the CT image. The best result reported in this work is Area Under the ROC Curve (AUC) of 0.61 and a binary accuracy of 61.5/100.

Authors: Hasib Zunair

To detect vehicles in bangladesh (i.e rickshaws, vans etc.) and extract the current road condition in the form on intensity. YOLO Object detection is to be used which will be trained on a custom dataset of the vehicles in the context of Bangladesh.

Design and Implementation of an Automated Web Based Multifunctional Attendance System

(Best Paper Honorable Mention)

Implementation of an automated multifunctional attendance system which uses RFID, FINGERPRINT, and real time facial recognition. Entry logging is displayed on a website.

Provides security and smooth data aquisition.

Authors: Hasib Zunair

(6th place overall)

A convolutional neural network architecture named VGG16 is used as a baseline to recognize bengali handwritten digits. Testing accuracy of 97.5% is achieved which led to the 6th position overall among 50 teams in the kaggle competition.

Design and Implementation of an IOT based Monitoring System for Inland Vessels using Multiple Sensor Network

The paper includes designing and implementing a wireless sensor network with a real time web application for monitoring multiple ships to prevent catastrophic events due to overloading.

The idea consists of four main parts: Detection Module, GPS tracker, communication system (NRF24L01+) and software application for web interface.

Authors: Hasib Zunair

This project demonstrates a hand gesture recognition system using skin color segmentation and image filtering techniques.

Design and Implementation of Security Patrol Robot using Andrioid Application

Here proposed, is a low-cost autonomous mobile security robot based on a multisensor system.The robot is mobile and provides fully autonomous patrolling in a defined area.

The system also provides four kinds of alarming systems via which notification of security violation is sent.

Authors: Aiman Snigdha, Tanzil Sharia, Hasib Zunair

An unmanned land rover was built to monitor different parameters in the mangrove forest for research purposes. The parameters included different gases and climate changes from surroundings.

(1st Position)

Authors: Hasib Zunair, Aiman Snigdha, Tanzil Sharia

An unmanned underwater robot was build for marine research and patrol purposes. In the demo, a depth of 2 feet was achieved.

Authors: Aiman Snigdha, Hasib Zunair, Tanzil Sharia, Abir Mahmud, Irfanul Haque, Shoumik Shekhar, Shahriar Kabir Rafi, Sajid Amin

A small unmanned ground vehicle (UGV) was developed as per requirements of the competition, the robot had the functions of moving autonomously across a storage area, counting items of a specific geometry in the warehouse, recording their ID tags, and carrying instrument payloads.

The robot obtained its surrounding visuals using a LIDAR Lite V3 and a regular camera. A 3D grayscale image is plotted with which the readings of the distance at every interval of time is measured and sent to a UI made in processing.

Authors: Tahzib Mashrik, Baized Shuvo, Hasib Zunair

(1st Position)

An autonomous wallfollower robot, which was the champion at Techfest '15 organized by Project Be, held at Dhaka, Bangladesh.

My teaching activities are shown here.

Conducted a technical day long workshop on fundamentals of computer programming using Python as the choice of language. This session was organized by Robotics and Automation Society IEEE NSU Student Branch Chapter. Materials are avaiable on github.

Conducted a hands on technical session on compter vision where participants where exposed to image manipulation, segmentation and feature extraction. The open source computer vision library(OpenCV) was used. Materials of the session can be found on github.

I was an instructor for a semester long workshop on robotics. The workshop started off with embedded electronics and programming and later on robotics.