Artificial Intelligence System for Analyzing Broiler Activity Index

• 5 min read
• Bookmark
• Main Blog Page

Broiler activities such as locomotion and movement can be quantified as the broiler activity index (BAI), which is automatically computed by determining changes in broiler-representing pixels between adjacent images. The BAI has been associated with leg health, productivity, and physical conditions of broilers; however, the indicator requires engineering knowledge to adjust image processing parameters, segment individual broiler chickens, select calculation areas, and normalize variations to obtain consistent values. The current calculation procedures are not user-friendly for producers or animal scientists without image processing experience.

The goal of a recently completed USPOULTRY-funded project was to develop a web-based artificial intelligence system for analyzing broiler activity index with three specific objectives: 1) verifying the biological meaning of the BAI; 2) exploring efficient algorithms to segment individual birds from images; and 3) developing a user-friendly interface for analyzing BAI.

The BAI was classified into high, medium, and low levels using machine learning models. High and medium levels of activity were significantly lower for broilers with cyclical heating operations than for those without heating operations. The results also indicated that the modified general deep learning model without extensive training can achieve satisfactory performance (>84% accuracy) in segmenting birds from poultry housing images. A user-friendly platform was developed using Streamlit to calculate the BAI, either individually or in groups, from videos. The platform was released into the GitHub repository with detailed steps for setup and operations.

The project summarizes that the BAI can be an effective indicator to indicate broiler heat stress, which can support early and timely interventions to improve broiler performance. The open-source, user-friendly platform allows researchers to interact with software tools to understand animal behavior patterns and welfare without needing extensive programming knowledge. Detailed quantitative measurements of poultry behaviors via the developed tool can better unravel treatment effects and further assist in producing broilers with better well-being and productivity. Company technicians can access and deploy the tool to develop low-cost commercial products for automatic monitoring, further improving the automation level of the poultry industry and reducing labor for flock inspection.

<Article Tags>

• Artificial Intelligence
• Broiler
• Machine Learning