AI model aims to identify chick sex at day 4 without opening eggs

A paper published April 30, 2026, in Animals reports a non-destructive, image-based method for identifying the sex of chicken embryos on day 4 of incubation using an improved MobileViT-V3 deep-learning model. The team from Huazhong Agricultural University said the model was designed to better detect weak embryonic blood-vessel patterns that are often obscured by shell texture and uneven lighting. In testing on a self-constructed dataset, the model reached 92.26% accuracy, 92.15% F1-score, 92.12% recall, and a kappa coefficient of 0.845, while running at 97.6 frames per second with 2.98 million parameters, which the authors said is fast enough for industrial sorting lines. (studocu.vn)

Why it matters: For veterinary professionals following poultry welfare and hatchery technology, the study adds to the push toward earlier in-ovo sexing as an alternative to culling day-old male chicks. The timing is important: the authors argue that day-4 identification could reduce resource use and better align with animal welfare expectations, while existing commercial systems often work later in incubation and may rely on fluid sampling or other more invasive workflows. Commercial players already market in-ovo systems based on DNA or optical methods, and adoption is expanding in Europe and the U.S., but those systems generally operate later than day 4 and often emphasize integration into hatchery operations over ultra-early detection. (studocu.vn)

What to watch: The next question is whether this day-4, vision-based approach can be validated across breeds, shell colors, hatchery conditions, and independent commercial trials before it can compete with more established in-ovo sexing platforms. (studocu.vn)