PhD, University at Buffalo, United States
Ayush Roy is an emerging researcher and innovator in the field of Electrical Engineering with a deep interest in AI, computer vision, and biomedical image analysis. Currently pursuing his B.E. at Jadavpur University, he has demonstrated exceptional potential through interdisciplinary research, AI-driven solutions, and impactful contributions to both academia and real-world applications. With multiple international publications and recognitions, Ayush is a dynamic force in the intersection of deep learning, signal processing, and intelligent systems.
Publication Profile
Google Scholar
🎓 Education Background
Ayush Roy is a final-year undergraduate student at Jadavpur University, West Bengal, India, enrolled in the Bachelor of Engineering (Electrical) program with an SGPA of 8.1/10 (2020–2024). He completed his schooling from Bhartiya Vidya Bhavan, West Bengal under the CBSE board, scoring 90.6% in Class 12 and a perfect CGPA of 10 in Class 10.
💼 Professional Experience
Ayush’s research journey began at Jadavpur University, working under renowned professors in Audio Signal Processing, Reinforcement Learning, and Image Segmentation. As a research intern at the Indian Statistical Institute, he contributed to dataset development and text detection models. He furthered his research as an intern at the University of Malaya on transformer-based networks and at IISc Bangalore on CLIP for image quality assessment. His work integrates deep learning models like YOLO, Swin Transformer, UNet, and CLIP with novel architectures and real-world applications.
🏆 Awards and Honors
Ayush has earned several accolades such as the Most Innovative Solution award at Hack-a-Web by NIT Bhopal (2021), 3rd Prize at FrostHack, IIT Mandi (2022), Top 10 in Cloud Community Hackday by GDG Cloud, and became a Finalist in both the IEEE R10 Robotics Competition and 404 Resolved hackathon at IIT Delhi.
🔬 Research Focus
His primary research areas include computer vision, medical image segmentation, scene text detection, and real-time AI systems. He is especially focused on lightweight models, attention mechanisms, domain adaptation, and hybrid approaches combining deep learning and signal processing. He has created multiple datasets for benchmarking including those for drone license plate detection, underwater text, water meter digit recognition, and circuit component recognition.
📌 Conclusion
Ayush Roy stands as a committed and creative researcher, blending electrical engineering fundamentals with cutting-edge AI methodologies. His work not only adds value to academic literature but also paves the way for practical, socially impactful AI systems. With an impressive early-career portfolio, Ayush continues to show immense promise for future contributions to science and technology.
📚 Top Publication Notes
AWGUNet: Attention-aided Wavelet Guided U-net for nuclei segmentation in histopathology images
Year: 2024
Journal/Conference: ISBI 2024
Cited By: 2 articles (Google Scholar)
A Wavelet Guided Attention Module for Skin Cancer Classification
Year: 2024
Journal/Conference: ISBI 2024
Cited By: 1 article (Google Scholar)
A New Lightweight Attention-based Model for Emotion Recognition Using Distorted Social Media Images
Year: 2023
Journal/Conference: ACPR 2023
Cited By: 3 articles
Fourier Feature-based CBAM and Vision Transformer for Text Detection in Drone Images
Year: 2023
Conference: ICDAR WML 2023
Cited By: 1 article
A Lightweight Script Independent Scene Text Style Transfer Network
Year: 2024
Journal: International Journal of Pattern Recognition and Artificial Intelligence (IJPRAI)
Cited By: 1 article
Identification and Classification of Human Mental Stress using Physiological Data
Year: 2022
Conference: IEEE CATCON 2022
Cited By: 4 articles
Adapting a Swin Transformer for License Plate Number and Text Detection in Drone Images
Year: 2023
Journal: Artificial Intelligence and Applications (AIA)
Cited By: 2 articles
An Attention-based Fusion of ResNet50 and InceptionV3 Model for Water Meter Digit Recognition
Year: 2023
Journal: Artificial Intelligence and Applications (AIA)
Cited By: 1 article
DAU-Net: Dual Attention-aided U-Net for Segmenting Tumor Region in Breast Ultrasound Images
Year: 2023
Journal: PLOS ONE
Cited By: 6 articles