GSoC 2025 Proposal: Deepfake Detection System using OpenCV

1

Hi OpenCV Community,

I’m Juslin Rafena R, a 3rd-year undergraduate student pursuing B.E. in Civil Engineering at Thiagarajar College of Engineering, Madurai, India, with an expected graduation in April 2027

I’m passionate about Artificial Intelligence and Computer Vision, and I’m excited to contribute to OpenCV through Google Summer of Code 2025. I’ve submitted my GSoC proposal titled:

Deepfake Detection System using OpenCV

In this project, I aim to build an efficient system to detect deepfakes using visual feature extraction, temporal inconsistencies, and machine learning, leveraging OpenCV’s tools and vision pipelines.

I would be truly grateful if any mentors or community members could review my proposal and provide feedback or suggestions for improvement.

:link: Link to Proposal PDF: Juslin Rafena_GSoc Project Proposal.pdf - Google Drive

Looking forward to learning and growing with OpenCV!

Warm regards,
Juslin Rafena R

2

hi, .. could not resist picking at your pdf there:

Deliverables

  • A Python-based deepfake detection module using OpenCV and pretrained models.

imo you should write c++ code
(this should add automatic python wrappers).
python-only code is usually not accepted.

  • Week 9-10: Build frontend UI using OpenCV’s highgui or integrate with Flask.

same problem.
also, library code should not contain UI dependancies (maybe put that into a sample, don’t assume, that there is a human ‘user’ even …)

  • Week 1-4: Build basic facial landmark detection and frame extraction using OpenCV.

we already have 3 landmarks models in opencv-contrib/face, also yunet. please take a look.

gl, otherwise !