CV

Researcher

TCS Reserach (formerly Tata Research, Developement and Design Centre)

I am working as a researcher in the Visual Computing and Embodied Intelligence Lab, where my work primarily focuses on creating synthetic world environments using generative models that visualize imagined scenes and objects that do not exist.

• Computer Vision, Computer Graphics, Image Processing, Deep Learning

Reviewer (TPC)

IEEE International Joint Conference on Neural Networks

I reviewed various papers for this conference and provided feedback on whether they should be accepted or not.

Doctor of Philosophy

Indian Institute Science

Robotics and Autonomous Systems

• Robotics, Computer Vision, Machine Learning

Bachelor's of Technology

Indian Institute of Engineering Science and Technology Shibpur

Information Technology

• Computer Science

Senior Secondary

Jadavpur Vidyapith

Pure Science

• Physics, Chemistry, Mathematics, Computer Science

Secondary

Jadavpur Vidyapith

General

• General

Anomalous activity detection for mobile surveillance robots

US Patent App. 18/473,595

InGenCo: Integrated In-Place 3D Scenario Generation and Collaboration

2025 IEEE International Symposium on Mixed and Augmented Reality

FAV3R: Fast and Accurate 3D VR Sketch to 3D Shape Retrieval

2025 IEEE International Symposium on Mixed and Augmented Reality

A transmission model based deep neural network for image dehazing

Multimedia Tools and Applications

Dehazing and vision enhancement: challenges and future scope

IET Intelligent Multimedia Processing and Computer Vision

A deep learning model to detect foggy images for vision enhancement

The Imaging Science Journal

Multi-feature based hazy image classification for vision enhancement

Procedia Computer Science

Anomalous activity detection from ego view camera of surveillance robots

IEEE International Joint Conference on Neural Networks

A Framework for Sex Identification, Accent and Emotion Recognition from Speech Samples

13th International Conference on Computing Communication and Networking Technologies

Improved Diagnosis on Low Resolution Medical Images

Medical images are important because they are used in a more sensitive field, namely the medical field. Raw data obtained directly from medical acquisition devices may provide a relatively poor representation of image quality. The primary goal of this project is to improve the features and characteristics of medical images in order to make improved diagnoses. Also it is extremely difficult to transfer high resolution images (e.g., MRI, CT) over low bandwidth. So the captured medical image is converted to low resolution format(how we converted) and then transmitted to the other end. During transmission, data can suffer loss due to various types of noises. Degraded low resolution image received in the receiver side need to be reconstructed. Image Enhancement (IE) algorithms are introduced for carrying out the requirements of converting received low resolution medical images to high resolution. We proposed a dual GAN architecture in our work to convert sensitive low resolution images to high resolution images that can be processed in the medical field. The model is composed of two GANs where the first GAN enhances the entire image and the second GAN enhances the region of interest in the image. The second discriminator has a novel loss function which focus particularly on the region of interest in modeical images. The work was able to surpass many of the SOTA scores by obtaining a Structural Similarity of 0.84.

• B.Tech Final Year Project

Cartoon Emotion Recognition

Social media platforms are widely used by individuals and organizations to express emotions, opinions, and ideas. These platforms generate vast amounts of data, which can be analyzed to gain insights into user behavior, preferences, and sentiment. Accurately classifying the sentiment of social media posts can provide valuable insights for businesses, individuals, and organizations to make informed decisions. To accomplish this task, a customized private cartoon dataset (original images) of social media posts has been provided, which contains labels for each post's emotion category, such as happy, angry, sad, or neutral. The task is to build and fine-tune a machine-learning model that accurately classifies social media posts into their corresponding emotion categories, using synthetic images. Where we obtained an accuracy of 95% on competition dataset.

• Hackathon Revelation 2023