Forensics

Inspiration

Crime scenes are often chaotic, with clutter, crowds, and missing details. Investigators need clear, accurate visuals — but manual cleanup is slow and risky. We wanted to build a tool that makes documentation faster, cleaner, and legally reliable.

What it does

Our app captures raw photos from a scene and uses FLUX.1 Kontext to:

🧹 De‑occlude and reconstruct hidden objects 📝 Auto‑annotate evidence (blood stains, weapons, traces) 🔒 Preserve originals + edits with metadata for chain‑of‑custody 📑 Export ready‑to‑use reports

How we built it

I used react and FLux.1 to built it.

Challenges we ran into

Finding open, ethical datasets for testing forensic scenarios

Accomplishments that we're proud of

Built a working prototype that processes and annotates images in second

What we learned

Context chaining in image models requires feeding outputs back in (no built‑in memory like LLMs)

What's next for Forensics

Make it more realistic

Built With

• flux.1
• react