About
Hello! I’m a researcher passionate about developing innovative solutions in artificial intelligence, computer vision, image processing, healthcare AI, robotics, and defence AI. My work bridges cutting-edge research with real-world applications, aiming to create technologies that address challenges in video analytics, medical diagnostics, human-AI interaction, and autonomous systems. I have presented my research at esteemed conferences, including ISBI (International Symposium on Biomedical Imaging), AIxPAC- AIxIA, IEEE GlobConET, IEEE INOCON, and NCVPRIPG, where I showcased optimized solutions for medical imaging, thermal imaging, computer vision, and more. I am honored to have received the Best Practical Paper Award at AIxPAC 2023 for my contributions to night-vision face detection using thermal imaging. I am committed to transforming AI research into practical solutions that can drive innovation across healthcare, robotics, and defence applications. If you share a passion for computer vision, healthcare AI, robotics, or defence AI, let’s connect and collaborate to shape the future of intelligent systems.
Working Areas
- Computer Vision
- Image Processing
- Machine/Deep Learning Optimization
- Human-Ai Interaction
- AI in Healthcare
- AI in Defence
What I've Done
Total Projects
Total Publications
Awards
Collaboration
Publication
"FESS Loss: Feature-Enhanced Spatial Segmentation Loss for Optimizing Medical Image Analysis." 2024 IEEE International Symposium on Biomedical Imaging (ISBI). Link
"Intelligent Virtual Worlds: A Survey of the Role of AI in the Metaverse." 2024 3rd International Conference for Innovation in Technology (INOCON). Link
"Seeing in the Dark: A Different Approach to Night Vision Face Detection with Thermal IR Images." AIxPAC under AIxIA 2023. Link
To Know More:
Projects
Generative AI Playground (Vision)
Building an interactive platform for experimenting with various GANs and VAEs, facilitating image tasks like data augmentation, image compression, image generation and reconstruction
Real-Time Depth Estimation & Image Reconstruction from Point Cloud
Developed a real-time depth estimation system that leverages the MiDaS model to generate accurate depth maps and YOLOv5 for object detection from webcam input.
Gaze Tracing for Mouse Pointer Movement
Implemented gaze-based mouse control, translating detected gaze direction into precise mouse pointer movements for improved user interaction and accessibility.
Medical Image Segmentation
Processes 3D medical images from various modalities, such as MRI and CT scans. Implemented segmentation in both centralized and decentralized setups with a novel loss function tailored for each approach.
City Space Segmentation
Developed semantic segmentation models for classifying and segmenting urban street scenes from the Cityspace dataset. And achieved 88-94% segmentation accuracy by optimizing model architectures and employing data augmentation techniques.
Low Light Night Vision Face Detection Using Thermal Imaging
Built a face detection module for low-light conditions using Haar-cascade, Dlib, and YOLOv8. And Enhanced detection accuracy by combining multiple models (e.g Dlib with YOLOv8) for improved performance in challenging lighting.
Palm Print Detection & Identification Using Mobile Cameras
Developed proprietary palm print dataset with manual annotation. Pre process through noise removal and kernel filtering. Applied transfer learning with YOLO (v5 and v8), with YOLOv8 outperforming YOLOv5 by 9-13%.
Predicting Solar Power Generation With Its Maintenance Activities
Leveraged deep learning, time series modelling, and various ML techniques & libraries (Sci-Kit Learn, Pandas, NumPy, TensorFlow, Keras, PyBats, matplotlib) to analyze data patterns, perform mathematical operations, and yield precise results.
Testimonials
Dr. Kshitij Jadhav
Assistant Professor, KCDH, IITB
Charul is a highly motivated innovative an enterprising individual who has worked closely with me on an important research project. He developed a novel Loss function for medical image segmentation. This helped in beating the SOTA results in low resource setting. Charul is a fantastic Python coder, works cohesively in a team and works deligently within deadlines. He is eager to learn and has the right attitude and a growth mindset which brings the best not only within himself but also his team.
Contact
Location:
Bharat (India)