Jay Nareshbhai Chaudhari
Visvesvaraya Research Scholar · IIIT VadodaraJay Nareshbhai Chaudhari is a Visvesvaraya Research Scholar at the Indian Institute of Information Technology (IIIT) Vadodara, working on Computer Vision & Edge AI. He holds a Master of Engineering in Electrical Engineering (Automatic Control & Robotics) from The Maharaja Sayajirao University of Baroda.
Beyond his research, he is an avid half marathon runner and enjoys playing the ukulele and singing Gujarati folk songs.
Awards & Achievements
-
Mixed Track Winner — Sclera Segmentation Challenge, IJCB 2025
Won for segmentation model "ShapeGAN-DLV3+" as detailed in Table 4, IJCB 2025 Proceedings.
-
2nd Position — WACV 2023 Pedestrian Attribute Recognition Challenge (Track 1)
Team "JaiC21", detailed in Table 5, WACV 2023 Proceedings.
-
IndoML 2022 Travel Grant
Presented a poster on Generalization Study on Person Attribute Recognition.
-
Shree Dewang Mehta IT Award — ₹1,00,000 Cash Prize (August 2019)
Awarded for Out-of-Syllabus Innovation Challenge by the Dewang Mehta Foundation Trust, Ahmedabad.
Research & Experience
- Designed and developed two comprehensive person retrieval systems:
- Single Query Discrete Framework: Developed novel CorPAR model — reduced parameters by 1.8–27% vs. SOTA with 1.63–8.07% F1 improvement. Achieved 22 FPS real-time on Nvidia Jetson Orin AGX via quantization & pruning.
- Multi-Query Discrete Framework: Designed DensePAR + Siamese network; applied SVD tensor decomposition achieving 20 FPS on edge device.
- Natural Language Queries: Designed and deployed a Transformer-based retrieval system for natural language queries.
- Developed AnnoTool — a Python GUI for annotating images and videos for object detection tasks.
- Developed QDP Studio — a unified model compression framework integrating quantization, pruning, decomposition, and knowledge distillation.
- Conducted hands-on tutorial sessions and mentored undergraduate students in computer vision, ML, and edge AI. GitHub: Tutorials
- Developed a two-stage deep learning system for individual mugger identification from UAV-captured data.
- Addressed challenges of scale variability, viewpoints, occlusions, and background clutter.
- Investigated system performance across data from two different seasons.
- Outperformed existing baselines by 4%.
- Developed a classification framework using Random Forest and Gradient Boosting Classifiers. GitHub: SSC
- Applied wavelet decomposition (Coiflet, Daubechies, Biorthogonal) for EEG feature extraction.
- Achieved 91% classification accuracy using Hjorth parameters.
Publications
J. N. Chaudhari, H. Galiyawala, P. Sharma, P. Shukla, M. S. Raval, "Onboard Person Retrieval System With Model Compression: A Case Study on Nvidia Jetson Orin AGX," IEEE Access, vol. 13, pp. 8257–8269.
J. N. Chaudhari, H. Galiyawala, M. Kuribayashi, P. Sharma, M. S. Raval, "Designing Practical End-to-End System for Soft Biometric-Based Person Retrieval From Surveillance Videos," IEEE Access, vol. 11, pp. 133640–133657.
H. Tripathi, J. N. Chaudhari, H. Galiyawala, P. Sharma, M. S. Raval, "Integrating Datasets with Discrete and Natural Language Annotations for Person Retrieval," IEEE VTC2023-Fall, Hong Kong, pp. 1–5.
J. N. Chaudhari, H. S. Dhiman, P. Suthar, K. Manjunath, "Wavelet Transform-Based Comparative Analysis of Wind Speed Forecasting Techniques," Renewable Energy Optimization, Planning and Control, Springer Singapore, pp. 121–128.
Inventors: Manish Kumar, Piyush Baldha, Jay Chaudhari, Sumit Upadhyay, Krunal Patel, Dr. Dipankar Deb | Published: 06/09/2019. Innovated a copper duct-based natural air ventilation system with water flow for temperature control in vehicles, enhancing passenger comfort and energy efficiency.
Services
I offer a range of academic and professional services to students, institutions, and organisations working in AI, computer vision, and related fields.
Hands-on Lab Sessions
Practical, project-based sessions on ML, deep learning, computer vision, and Edge AI — tailored for university students and researchers.
Mentorship
One-on-one guidance for students navigating research careers, thesis projects, publication writing, and PhD applications in AI/ML.
Talks & Lectures
Invited talks on surveillance systems, model compression, biometric recognition, and real-time AI at the edge.
Consultancy
Technical consultancy for organisations seeking to deploy AI-powered surveillance or edge-optimised vision pipelines in production.
Interested in working together? Feel free to reach out.
Get in TouchProfessional Services
Conferences
Journals
- Mar 2024Deep Dive into Dimensionality Reduction: A Machine Learning Approach — LDRP University
- Jul 2023Deep Learning: Concepts to Deployment — Nirma University
- May 2023U-Net and Other Models for Image Segmentation — Parul University
- Local Organising Committee — National Conference on GeneRative AI for Nurturing Sustainable Agriculture (GRAINS 2024), October 2024.
- Volunteer — 18th IEEE International Conference on Vehicular Electronics and Safety (IEEE ICVES), December 2024.
- Volunteer — IEEE ITSS Two-Wheeler Safety & Mobility Technologies Awareness Campaign, Ahmedabad University, November 2023.