Akshay Daydar

Akshay Daydar

Akshay Daydar is a PhD candidate in the Department of Mechanical Engineering at the Indian Institute of Technology Guwahati (IIT Guwahati). His research interests include biomedical image processing, deep learning, and human gait analysis. Prior to joining IIT Guwahati, Akshay completed his postgraduate and undergraduate studies at the National Institute of Technology Warangal (NITW) and Government College of Engineering, Amravati (MH) respectively. He has also interned at BHEL Corporate R&D Hyderabad for his postgraduation project. Besides his academic pursuits, he enjoys photography, sketching, and traveling.

Contact Details

Downloads

Experience

Bharat Heavy Electricals Limited (BHEL)

July 2019 - May 2020

Intern at Department of Experimental Mechanics and Virtual Reality (EMV), R&D Hyderabad

Supervisors: Prof. N. Selvaraj (ME, NITW) and Mr. Vimal Kumar Gaurav (Senior Engineer, BHEL)

Project Title: Predictive Modelling of the Steam turbine generator unit and critical parameters Identification for rotor vibration and bearing problems.

Description: This project used data-driven modeling to predict steam turbine and generator outputs and identify critical parameters for rotor vibration and bearing issues. Techniques included ANN, RNN, Decision Trees, and Random Forests with feature selection via PCA and Pearson correlation. The final model suggested critical parameters and included a GUI for visualization and dynamic analysis.

Keywords: Predictive Modelling | Statistical Data Analysis | Classical Machine Learning Models | Explainability Methods. Software: R (Interfaces: R studio, R-Shiny).

Publication:

1. Vimal Kumar Gaurav, Daydar Akshay, N. Selvaraj, Krishnaiah Jallu, K. Ramakrishna. "Comparison of Predictive Modeling Techniques for Determining Power Output in The Thermal Power Plant", International conference on machine learning and data Analytics 2020 (Best Paper Award).

Ph.D. Project

September 2020 - Ongoing

Supervisors: Prof. S. Kanagaraj (ME), Prof. Arijit Sur (CSE)

Project Title: Classification of Knee Osteoarthritis from Radiological and Gait Data Using Deep Learning Models and Multimodal Approach.

Description: This study focuses on the development of automated tools for the comprehensive analysis of Knee Osteoarthritis (KOA). The key objectives include: 1. Automatic classification of KOA, 2. Segmentation of tissues involved in its pathophysiology, 3. Extraction and analysis of gait parameters from KOA patients, and 4. Establishing correlations between radiological findings and gait abnormalities. The entire workflow is automated using deep learning techniques, with medical validation provided by clinical experts. This interdisciplinary research lies at the intersection of deep learning, biomechanics, and clinical sciences. The radiological data used in the study include X-ray and MRI scans, while gait data are collected from the Gait and Motion Analysis Lab at IIT Guwahati. Overall, the study delivers a novel multimodal framework that integrates radiological and biomechanical data, enabling accurate classification and segmentation, as well as generating a radiological–gait correlation matrix. This approach aims to reveal the intrinsic interdependence between multiple diagnostic modalities for a deeper understanding of KOA.

Keywords: Semantic Segmentation | Multiclass Classification | Multilabel Classification | Gait Data Collection | Medical Context Aware Deep Learning Models. Software: Python (Interfaces: PyTorch, Monai, TensorFlow).

Publications

1. "Comparison of Predictive Modeling Techniques for Determining Power Output in The Thermal Power Plant" – ICMLDA 2020 (Best Paper Award)
2. "Diffusion Based Shape-Aware Learning with Multi-Scale Context for Segmentation of Tibiofemoral Knee Joint Tissues: An End-to-End Approach." – Accepted at IEEE ICIP 2025 [Preprint]
3. "DeepOsteoCLS: Deep Learning-Based Framework for Knee Osteoarthritis Classification With Visual Explanations from Radiographs and MRI Volumes." – Under Review at Biomedical Signal Processing and Control (Elsevier, Q1)
4. "Med-SeAM: Medical Context Aware Self-Supervised Learning Framework for Anomaly Classification in Knee MRI." – Accepted at ACM ICVGIP 2024[Link]
5. "MedCAM-OsteoCls: Medical Context Aware Multimodal Classification of Knee Osteoarthritis." – Accepted at IEEE ICASSP 2025[Link]
6. "Development of a Portable Plantar Image Acquisition Setup and its Analysis for Pressure Distribution and Arch Index of Varied Foot-types using an Artificial Intelligence." – Accepted at IEEE CONECCT 2025
7. "Automated Flat Foot Diagnosis using Generative Model and Color Channel Feature Integration Network." – Accepted at IEEE CONECCT 2025
8. "River-GEM: Generating and Enhancing Muddy Water Images." – Accepted at IEEE ICASSP 2025[Link] Co-authored

Teaching Assistant for Deep Learning (CS590) and Mechanical Workshop (ME101)

Technical Assistant for Universal Testing Machine (UTM) – 5KN at Central Instrument Facility at IITG

Departmental Placement Representative (DPR) for Ph.D. in 2023-24.

Volunteered and conducted 1 lecture and 1 hands-on session at Winter School on Deep Learning for Vision and Language Modelling 2025, IITG

Volunteered and conducted 1 lecture and 1 hands-on session at Summer School on Language and Vision Modelling (LAVA 25), IITG

Reviewer of: