A Method for Automated Detection of Chicken Coccidia in Vaccine Environments

Area of Science:

• Veterinary Parasitology
• Artificial Intelligence in Animal Health
• Poultry Science

Background:

• Chicken coccidiosis causes significant economic losses in the poultry industry.
• Accurate detection of Coccidia oocysts is vital for vaccine quality control and efficacy.
• Existing detection methods face challenges due to oocyst size, orientation, and species similarity.

Purpose of the Study:

• To develop an accurate and efficient model for detecting chicken Coccidia oocysts in vaccine samples.
• To improve the automated assessment of vaccine quality and enhance animal welfare.

Main Methods:

• Proposed YOLO-Cocci, a chicken coccidia detection model based on YOLOv8n.
• Incorporated an efficient multi-scale attention (EMA) module in the backbone.
• Developed an inception-style multi-scale fusion pyramid network (IMFPN) for the neck.
• Designed a lightweight feature-reconstructed and partially decoupled detection head (LFPD-Head).

Main Results:

• YOLO-Cocci achieved an mAP@0.5 of 89.6%, a 6.5% improvement over the baseline.
• Model parameters and computation were reduced by 14% and 12%, respectively.
• mAP@0.5 for Eimeria necatrix detection increased by 14%.

Conclusions:

• YOLO-Cocci significantly enhances the accuracy of chicken coccidia oocyst detection in vaccines.
• The model offers improved efficiency and reduced computational cost for automated quality assessment.
• This technology contributes to better vaccine quality, reduced economic losses, and improved animal welfare in the poultry industry.