Abhinav Raundhal

Abhinav Raundhal

IIIT Hyderabad

Hi! I am an Undergraduate Researcher at the Center for Visual Information Technology (CVIT), advised by Prof Ravi Kiran Sarvadevabhatla. My research areas include 3D computer vision and document analysis. I am also an undergrad researcher at the Robotics Research Center (RRC) at IIIT Hyderabad, advised by Prof Madhav Krishna. I have recently worked on projects on improving reconstruction quality in Gaussian Splatting, autonomous exploration in 3D environments, and text detection in complex handwritten documents. I am always ready to learn. In my free time, I like to travel, paint, and do photography.

Abhinav Raundhal

News

Nov 2025
Our paper "STRinGS: Selective Text Refinement in Gaussian Splatting" was accepted to WACV 2026!
Jan 2025
Joined the Robotics Research Center (RRC) as an Undergraduate Researcher.
May 2024
Joined the Center for Visual Information Technology (CVIT) as an Undergraduate Researcher.

Research Publications

STRinGS: Selective Text Refinement in Gaussian Splatting

Abhinav Raundhal*, Gaurav Behera*, P. J. Narayanan, Ravi Kiran Sarvadevabhatla, Makarand Tapaswi

WACV 2026

Developed a text-aware selective refinement framework that significantly improves fine-grained text reconstruction in 3DGS, achieving a 63.6% OCR-CER improvement over baseline methods.

Project Page Code

Projects

DistLearn: Federated Learning

Built a federated learning system enabling multiple clients to train locally with data privacy using methods like FedSGD, FedAvg, FedAdp with SSL/TLS encryption.

Distributed SystemsPrivacy-Preserving ML
Code

6DoF Pose Estimation

Developed a framework for 6-DoF pose estimation from meshes using differentiable rendering techniques, optimizing pose parameters.

MeshesGraphics
Code

Image Editing with NNF

Implemented PatchMatch for tasks like in-painting, watermark, and object removal leveraging classical image processing.

Image Processing
Code

Paddy Procurement

Used image processing techniques and ML for crop identification, detecting fake paddy, and determining health stages.

Computer VisionML
Code

Car Insurance Assessment

Built a low-latency pipeline for claim assessment, incorporating fake image detection, blur checks, and text-image consistency.

Computer VisionML
Code

Network File System

Developed a network file system consisting of a naming server, storage servers and clients, with functionalities like reading, writing, creating, copying, deleting files & directories; handling multiple client requests simultaneously.

Computer VisionML
Code