Rami Ayouni
  • Enactus ENET'Com
  • Team project
  • Jan 2025 – present

TUT'ECT: early detection of Tuta absoluta

~4,000

self-annotated images · 2 national awards

~4,000

~4,000 self-annotated images

  • How it was measuredImages labelled by the team for object detection
  • On what~4,000 images
  • Keep in mindNo published evaluation metrics

Enactus ENET'Com team · Rami: Project Manager

Watch the demo

The problem

TUT'ECT: early detection of Tuta absoluta · before

  • A fast pest

    Tuta absoluta larvae tunnel through tomato leaves and fruit, and an infestation spreads fast.

  • Costly spraying

    The usual answer is repeated insecticide spraying: expensive and harmful.

  • Sprayed blindly

    It is often sprayed on a schedule rather than when it's needed.

The goal

TUT'ECT: early detection of Tuta absoluta · after

  • Early detection

    Catch the first signs early from a simple leaf photo.

  • A natural fix

    Point the farmer to a natural alternative, Trichogramma, a tiny wasp that destroys the moth's eggs.

How it works

TUT'ECT: early detection of Tuta absoluta

Leaf photo YOLO detector Verdict Healthyno action TutaTrichogramma
  1. Step 1 of 4 The farmer signs in and uploads a leaf photo from the field.
  2. Step 2 of 4 The YOLO model, trained on ~4,000 images the team annotated by hand, draws a box around each leaf region with a label ("Healthy" or "tuta absoluta") and a confidence score.
  3. Step 3 of 4 No infestation found: "No Tuta detected — your crops look healthy."
  4. Step 4 of 4 Infestation found: the infested areas are shown, and the app recommends Trichogramma as a natural treatment, with a way to order it.

Next project

7 % → 81 % CBAM compliance agent ← Previous: JobMatch: LLM-assisted candidate pre-selection

Sources

  1. ~4,000

    ~4,000 self-annotated images

    • How it was measuredImages labelled by the team for object detection
    • On what~4,000 images
    • Keep in mindNo published evaluation metrics