KT-YOLO: A multi-convolution kernel collaboration model for dense Hu sheep behavior detection

Suoxiang Zhang, Hongrui Chang, Zhonghong Wu, Guoxin Wu, Ronghua Ji · 2026

Read KT-YOLO: A multi-convolution kernel collaboration model for dense Hu sheep behavior detection by Suoxiang Zhang, Hongrui Chang, Zhonghong Wu, Guoxin Wu, Ronghua Ji (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0349267.txt.

Description

Suoxiang Zhang and colleagues built a camera system that watches crowded pens of Hu sheep and sorts out what each animal is doing. Dense groups make this hard, since sheep block each other from view and common actions swamp rare ones in training data. The team first gathered a new image set showing thousands of sheep standing, lying, eating, and drinking on intensive farms.

Their model, KT-YOLO, reads each image at several scales at once so partly hidden animals still register, and it uses a training tweak that pays extra attention to rare actions. It beat the standard model it grew from by a clear margin and also outperformed a newer rival. The authors aim at round-the-clock welfare checks on busy farms without extra staff.

Catalog record

CREATORSuoxiang Zhang, Hongrui Chang, Zhonghong Wu, Guoxin Wu, Ronghua Ji
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE47912
LICENSECC_BY
RIGHTSCC BY
SOURCEPLOS

Source: https://doi.org/10.1371/journal.pone.0349267

Browse the full library · About the Tower of Babel