Quail Egg Detection and Classification Through Egg Candling and Image Processing

Clients

Techstack

Purpose

Checking whether a quail egg is fertile means candling it by hand — shining light through each shell and judging it by eye, one at a time. It's slow, subjective, and hard to scale for a hatchery. A thesis team wanted to prove a machine could make that call automatically and consistently.

Description

We built a candling device that backlights each egg, captures the image, and grades it — fertile, infertile, or not a quail egg at all — using a deep-learning vision model trained on real candling data and run directly on the device. A touchscreen interface handles capture and shows the verdict on the spot, and we validated the model's accuracy with a confusion matrix. It turns a tedious eye-strain task into a one-touch read — the kind of on-device computer-vision problem we take on and ship.