In a world where AI-generated images, filters and editing tools are making reality blurry — TruthLens builds a system where authenticity can be proven, not assumed.
Hi — I'm Gouri Krishna, a first-year B.Tech Cybersecurity student. This is my first serious project, and it's focused on one question:
"How do we know if what we're seeing is real?"
TruthLens is my attempt to answer that — starting small, building on solid foundations, and growing into something that matters.
This is just the beginning. More ideas are coming.
Every image processed by TruthLens goes through a multi-layer integrity system.
Every image gets a unique cryptographic fingerprint. Any pixel-level change produces a completely different hash — making tampering mathematically detectable.
LAYER 1 · INTEGRITYRSA/ECDSA signatures bind each capture to a specific device and identity. No one can fake an image as coming from you without your private key.
LAYER 2 · IDENTITYUsing LSB steganography, a unique ID and reference hash are hidden inside the pixel data — invisible to the human eye, readable by TruthLens.
LAYER 3 · TRACEABILITYA visible "TruthLens Verified" overlay provides public-facing proof and awareness that an image passed through the system.
LAYER 4 · AWARENESSEvery capture stores timestamp, device ID, GPS coordinates (optional), and chain-of-custody history. The full story of an image, locked in.
LAYER 5 · HISTORYUpload any image and get a verdict in seconds — AUTHENTIC, TAMPERED, or UNVERIFIED. No accounts, no delays, no ambiguity.
LAYER 6 · VERIFYUse your device camera directly in the browser or upload an existing image. Both paths feed into the same verification pipeline.
A SHA-256 hash is computed from the raw image bytes. This 64-character fingerprint is unique to the exact pixel data — one changed byte produces a completely different hash.
The system attempts to extract an invisible LSB watermark from the image. If found, it retrieves the original record from the registry.
The current hash is compared to the stored original. Signature validity is checked. The result is definitive.
Full details returned: status, hash value, timestamp, source, device ID, and chain-of-custody history.
Choose how you want to provide the image. Both options run through the same verification engine.
These mockups show the intended mobile application — in development as a Flutter app.
DCT-based frequency domain watermarks that survive compression, resizing, and filtering.
SOONFull RSA-2048 / ECDSA asymmetric signing to prove origin with mathematical certainty.
SOONHash values stored on a public blockchain — immutable, decentralized proof of existence.
PLANNEDML classifier to flag AI-generated or face-swapped images even without a watermark.
PLANNEDTrack every share, edit attempt, and verification event across a distributed audit trail.
PLANNEDAnyone can verify an image via a shareable URL — no app required to check authenticity.
PLANNEDThis project is independent, open to ideas, feedback, and collaboration. If this problem interests you — whether you're a developer, designer, researcher, or just curious — I'd love to hear from you.