EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Enhanced Methodology and Experimental Research for Caged Chicken Counting Based on YOLOv8

Zhenlong Wu iD, Jikang Yang iD, Hengyuan Zhang, Cheng Fang iD

Animals · 16.03.2025 · 15 (6) · 853

DOI: 10.3390/ani15060853 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Animals, online erschienen am 16.03.2025. Verfasst von Zhenlong Wu und 3 weiteren Autor:innen. Beteiligte Einrichtung: College of Engineering, South China Agricultural University, Guangzhou 510642, China. Die Arbeit wurde bisher 14-mal zitiert (Crossref) und nennt 43 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Guangdong Basic and Applied Basic Research Foundation, Project of State Key Laboratory of Swine and Poultry Breeding Industry, Guangzhou Science and Technology Planning Project.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurately counting chickens in densely packed cages is a major challenge in large-scale poultry farms. Traditional manual counting methods are labor-intensive, costly, and prone to errors due to worker fatigue. Furthermore, current deep learning models often struggle with accuracy in caged environments because they are not well-equipped to handle occlusions. In response, we propose the You Only Look Once-Chicken Counting Algorithm (YOLO-CCA). YOLO-CCA improves the YOLOv8-small model by integrating the CoordAttention mechanism and the Reversible Column Networks backbone. This enhancement improved the YOLOv8-small model’s F1 score to 96.7% (+3%) and average precision50:95 to 80.6% (+2.8%). Additionally, we developed a threshold-based continuous frame inspection method that records the maximum number of chickens per cage with corresponding timestamps. The data are stored in a cloud database for reliable tracking during robotic inspections. The experiments were conducted in an actual poultry farming environment, involving 80 cages with a total of 493 chickens, and showed that YOLO-CCA raised the chicken recognition rate to 90.9% (+13.2%). When deployed on a Jetson AGX Orin industrial computer using TensorRT, the detection speed increased to 90.9 FPS (+57.6 FPS), although the recognition rate slightly decreased to 93.2% (−2.9%). In summary, YOLO-CCA reduces labor costs, improves counting efficiency, and supports intelligent poultry farming transformation.

Schlagwörter

Stichwörter der Autor:innen

chicken countinglayered cage farmingobject detectionpoultryprecision livestock farming

Autor:innen und Institutionen

  • College of Engineering, South China Agricultural University, Guangzhou 510642, China.
  • Faculty of Bioscience Engineering, Katholieke Universiteit Leuven (KU LEUVEN), Kasteelpark Arenberg 30, 3001 Leuven, Belgium.
  • State Key Laboratory of Livestock and Poultry Breeding, South China Agricultural University, Guangzhou 510642, China.

ORCID-Profile: Zhenlong Wu, Jikang Yang, Cheng Fang

Förderung

  • Guangdong Basic and Applied Basic Research Foundation
  • Project of State Key Laboratory of Swine and Poultry Breeding Industry
  • Guangzhou Science and Technology Planning Project

Zitieren

APA

Wu, Z., Yang, J., Zhang, H., & Fang, C. (2025). Enhanced Methodology and Experimental Research for Caged Chicken Counting Based on YOLOv8. Animals, 15(6), 853. https://doi.org/10.3390/ani15060853

Vancouver

Wu Z, Yang J, Zhang H, Fang C. Enhanced Methodology and Experimental Research for Caged Chicken Counting Based on YOLOv8. Animals. 2025;15(6):853. doi:10.3390/ani15060853

BibTeX
@article{Wu2025,
  author = {Wu, Zhenlong and Yang, Jikang and Zhang, Hengyuan and Fang, Cheng},
  title = {Enhanced Methodology and Experimental Research for Caged Chicken Counting Based on YOLOv8},
  journal = {Animals},
  year = {2025},
  volume = {15},
  number = {6},
  pages = {853},
  doi = {10.3390/ani15060853}
}
Kostenlos starten →