Projects
Ongoing...
Research:
-
Few Shots Learning for Object Detection and Image Segmentation
-
Semi-Supervised approaches for 3D data.
-
Unsupervised Image Classification
-
Anomaly Detection
Implementation: Developed, Trained and Evaluated state of the art deep learning models for different computer
vision tasks
Tools and Libraries: Pytorch, Tensorflow, Pytorch-Lightning and MLflow

Associated with mindtrace(dot)ai
Data Science and Machine Learning Course
(Sept 2022 - Jan 2023)
Associated with Massachusetts Institute of Technology (MIT)

Projects:
-
Netflix Movie Recommendation Link of Certificate
-
Hotel Booking Cancellation Prediction Link of Portfolio
-
Pima Indians Diabetes Analysis
Lectures on topics: Foundations of Data Science, Statistics, Unstructured Data, Regression, Classification, Prediction, Deep Learning, Recommendation Systems, Networking and Graphical Models.
Tools and Libraries: Python, Pytorch, Pandas, SciPy
CVPR 2021 RetailVision AliProducts Challenge
March-June 2021
Associated with mindtrace(dot)ai
Team placed Rank 11th.

Detecting Seen/Unseen Concepts Online while Reducing Response Time with/without Bounding Boxes using Domain Adaptive Multimedia Event Processing
PhD Research Scholar
April 2016 - March 2021
Thesis Title: Detecting Seen/Unseen Concepts Online while Reducing Response Time with/without Bounding Boxes using Domain Adaptive Multimedia Event Processing

Summary: Deep neural network-based techniques are effective for image classification, but the limitation of having to train classifiers for unseen concepts may increase the overall response-time for multimedia-based event processing models. This work focuses on foundational aspects of the problem of reducing response-time for online adaptive classifiers-based multimedia event processing which includes introducing object detection operators, standardization of the concept of response-time, identification, and proposed multiple IoMT based deep neural network models while using object detection specifically You Only Look Once (YOLO), Single Shot MultiBox Detector (SSD), and RetinaNet, and applying transfer learning. Lastly, I report the best possible performance of current object detection models for the online construction of classifiers. The major challenge in training deep neural network-based models is the need to collect many images with bounding box annotations, which is impossible for millions of unseen concepts. My final specific work is the design of first and fast detector for the training of unseen classes using only image-level labels with no bounding box annotations. It takes 10 min only to train an object detector.
Languages: Python, C, Shell Scripting (Linux Platform)
Libraries: TensforFlow, CUDA, cuDNN, Keras, OpenCV
Hardware Used: Nvidia Titan Xp GPU
Publication Outcomes: 3 Journals, 4 Conference papers (in CVPR 2022 selected for Oral Presentation, Journals of IEEE Access, Elsevier, and Springer; and other conferences including ICMR and DEBS)
Image Segmentation Using Fuzzy Multi Criteria Decision Making
Masters Dissertation
July 2013 - June 2015

Title: Image Segmentation using Fuzzy Multi-Criteria Decision Making
Summary: Image segmentation refers to the separation of objects from the background. Practically it is impossible to design a segmentation algorithm that has 100% accuracy. In this dissertation, two methods of segmentation are proposed: the first one is the Improved Sobel Edge Detection algorithm and the second is the Falling Ball algorithm. Our Falling ball algorithm which is a region-based segmentation algorithm, an alternative to watershed transform (based on waterfall model) and applies Fuzzy Logic for the segmentation. Simulation results show that the proposed algorithms give superior performance over conventional Sobel edge detection methods and watershed segmentation algorithm.
Languages: C, Java, Shell Scripting (Linux Platform)
Publication Outcomes: 1 Journal with +150 citations
Edge Detection using Ant Colony Optimization
Masters Project (6 months)
May 2014 - June 2015

Title: Edge Detection using Ant Colony Optimization
Summary: In this work, a multi-threading-based implementation of Ant Colony Optimization (ACO) is proposed for identifying edges in images. It combines multi-threading with ACO for increasing the randomness among the artificial ants. Simulation results show that the proposed method has significantly lower execution time as compared to conventional ACO for edge detection.
Languages: C, Java, Shell Scripting (Linux Platform)
Libraries: POSIX
Publication Outcomes: 1 Conference paper
Framework development and implementation of stereoscopic website
Bachelors Project (6 months)
Dec 2012 - May 2013

Title: Framework development and implementation of stereoscopic website
Summary: In this work, we worked on 3D images for the development of a Stereoscopic Website. We analyzed MPO and anaglyph 3D image formats. Moreover, we presented a new algorithm for obtaining depth information (for Depth-Map) pertaining to a depicted scene from a set of available pair of stereoscopic images.
Languages: C and HTML (Linux and Windows Platform)
Hardware Used: 3D television and stereoscopic glasses
Publication Outcomes: 1 arxiv paper and Book “Towards Stereoscopic Websites”
Simulation of M/G/1 Queue
Bachelors Project (6 months)
July 2011 - Dec 2011

Languages: C, Java, Shell Scripting
Text to Speech Converter
Bachelors Project (3 months)
Oct 2010 - Dec 2010

Languages: C and Java