── ABOUT ──

Four students.
One detector.

SeeThrough is a capstone project by four software engineering students at the University of Ottawa — built to help journalists and moderation teams identify AI-generated video content before it spreads.

── THE STORY ──
01 / Problem

Even seasoned journalists and moderators miss convincing AI videos -- multiple reviewers still let synthetic clips slip through.

02 / Approach

We built a proprietary pipeline that extracts deep visual representations from video frames and classifies them using a trained binary model.

03 / Result

Strong accuracy on real compressed social content -- with confidence scores your team can act on.

── TECH SPEC ──
Vision
Transformer Backbone
deep semantic feature extraction
Classifier
Discriminative Model
margin-based binary classification
Inference
Optimised Runtime
half-precision acceleration
API
FastAPI
async REST endpoints
Frontend
Next.js 16
App Router + Tailwind v4
Anim.
GSAP
scroll and transition layer
── TEAM ──

Meet the builders.

YA

Yusuf Al-Khawaldeh

Backend / DevOps

yalkh091@uottawa.ca

AA

Asad Ali

Backend / API

mali229@uottawa.ca

BL

Bryan Luu

ML / Frontend

bluu072@uottawa.ca

MY

Mohamed Yatera

ML / Research

myate081@uottawa.ca