GSoC 2025 Proposal – ViT-Based Face Recognition Module for OpenCV – Apara Chaurasia

[Apara Chaurasia]

Dear OpenCV GSoC Team,

I hope you're doing well.

My name is Apara Chaurasia, and I’m a third-year B.Tech Computer Science and Engineering student at SRM Institute of Science and Technology, Chennai, India.

I have submitted my GSoC 2025 proposal titled “ViT-Based Face Recognition Module for OpenCV.” This project aims to modernize OpenCV’s face recognition capabilities by integrating a Vision Transformer (ViT) pipeline. My deliverables include:

- A modular ViT-based face recognition model compatible with OpenCV’s existing pipeline
- Real-time performance optimization and CUDA backend support
- Enhanced robustness to pose, lighting, and occlusion
- Dataset support, benchmarking, and testing tools
- Visual demos and developer documentation

This proposal builds on my practical experience and academic exposure:
- At IIT Indore, I worked on malware detection using Hybrid ViT-CNNs  
- At IIT Mandi, I explored self-supervised learning with Vision Transformers (I-JEPA)  
- At DRDO, I built metallurgical classification models   

I’m passionate about contributing to OpenCV this summer and beyond. I would be very grateful for any feedback or guidance from the mentors regarding my proposal.
 
Apara Chaurasia