Vidhan Jain

Vidhan Jain

DL/ML Research Engineer

Master's student at NIT Rourkela specializing in Computer Vision, Thermal Imaging, and Deep Learning. Building intelligent systems that push the boundaries of visual perception.

View My Work Get In Touch

Experience

Research Intern

Space Application Centre, ISRO

June 2022 – November 2022

Ahmedabad, Gujarat

Under Anand S. Sahadevan

  • Preprocessed and structured large-scale airborne LiDAR datasets for comprehensive analysis
  • Developed an innovative method for identifying individual tree crowns from airborne LiDAR data
  • Implemented tree crown delineation using region-growing algorithms after identifying tree tops in rasterized middle and top layers of point clouds
  • Validated accuracy through comparison between estimated values and ground truth data

Research & Publications

Conferences

TSRNet: A Novel Lightweight Thermal Saliency Refinement Network for Tiny Human Detection

National Conference on Communications (NCC 2026) Accepted

Research Under Dr. Sobhan Kanti Dhara, NIT Rourkela

  • Proposed TSRNet, a lightweight saliency network achieving 0.454 mAP@0.5 on RGBT TinyPerson, outperforming SOTA detectors by 5.5% with only 36.5 GFLOPs.
  • Engineered Dual-Frequency Refinement (DFR) and Axial Context Pooling (ACP) modules to enhance weak thermal signatures via 3D contextual filtering.

SAPNet: A Lightweight and Efficient Saliency-Aware Path Network for Tiny Human Detection in Thermal Aerial Imagery

IEEE SPACE 2026 Under Review

Research Under Dr. Sobhan Kanti Dhara, NIT Rourkela

  • Developed SAPNet, achieving 0.404 mAP@0.5 with only 3.4M parameters and 20 GFLOPs, optimized for resource-constrained aerospace platforms.
  • Engineered novel Saliency-Path and Channel-Reweighting (SPCR) and S3-CSP modules to preserve subtle thermal signatures and enhance multi-scale feature aggregation via dynamic path allocation.

Journals

From Pixel To People: Semantic and Spatial Aggregation and Detection Network for Tiny Object Detection in Aerial Thermal Imagery

IEEE Geoscience and Remote Sensing Letters (GRSL) To be Submitted

Research Under Dr. Sobhan Kanti Dhara, NIT Rourkela

  • Architected the Center-Surround Weighted Channel (CSWC) Module to redistribute channel weights and enhance thermal signatures across three down-sampling layers, effectively capturing spatial context.
  • Engineered Contextual Wrapper Attention (CWA) to model global dependencies and contextual relations between object instances, significantly improving localisation accuracy.
  • Developed the Semantic-Spatial Fusion (SSF) Module in the network neck for high-level multi-scale feature fusion, specifically addressing feature misalignment to achieve superior results across 3 benchmark datasets.

Featured Projects

Eye Disease Classification & Model Optimization

Conducted comparative evaluation between ResNet50 and Vision Transformer for medical imaging, achieving 84.2% accuracy. Engineered inference optimization pipeline with ONNX and TensorRT, containerized using Docker for scalable cloud deployment.

Python PyTorch TensorRT Docker ONNX
View on GitHub →

Real-Time Face Mask Detection

Fine-tuned MobileNetV2 for edge deployment with low computational cost. Implemented post-training quantization and graph optimization via TensorRT. Deployed as lightweight Docker container for seamless integration with video surveillance.

Python MobileNetV2 TensorRT Docker OpenCV
View on GitHub →

Text to Image Generator

Built web application enabling image generation from text prompts using Stable Diffusion XL via Hugging Face Inference API. Deployed using StreamLit for intuitive user experience.

Python HuggingFace API Stable Diffusion StreamLit
View on GitHub →

Technical Skills

Languages

Python C C++ MATLAB SQL

ML/DL Libraries

PyTorch TensorFlow NumPy Pandas Matplotlib Sklearn Skimage OpenCV

MLOps & Tools

ONNX TensorRT Docker GitHub VS Code

Education

Master of Technology

National Institute of Technology, Rourkela

Aug 2024 – Present

Rourkela, Odisha

  • Specialization: Signal and Image Processing
  • Current CGPA: 9.69
  • Relevant Coursework: Computer Vision, Machine Intelligence, Soft Computing, Advanced DSP, Optimization Techniques, Linear Algebra, Probability, Multimedia Signal Processing

Bachelor of Technology

Jabalpur Engineering College

Aug 2019 – May 2023

Jabalpur, Madhya Pradesh

  • Major: Electronics and Communication Engineering
  • CGPA: 8.23
  • Foundation: Data Structures, Programming in C, Digital Signal Processing

Let's Connect

I'm always interested in hearing about new opportunities, research collaborations, or just chatting about AI and computer vision!

Email

vidhanjain2019@gmail.com

Phone

+91 97552 48639

LinkedIn

linkedin.com/in/vidhanjain

GitHub

github.com/vidhanjain03