Obtaining broiler chickens’ weight through depth image

Chicken is one of the most consumed proteins in the world. In order to efficiently supply this demand, the rearing system has been intensified and a modern broiler facility holds approximately 20,000 birds with a single caretaker walks the building to assess birds daily. Remote monitoring tools, such as depth cameras, can provide continuous monitoring to improve animal caretaking and provide accurate information for body development of the broilers. This research aims to automatically obtain the bodyweight of broilers through approximating the body dimensions using depth image. The Azure® Kinect depth sensor was used for image collection at top view position above the weighing scale and 10 images were collected from individual broiler. Data analysis and subsequent prediction of body dimensions were performed using an algorithm developed by MATLAB® (R2018a) to estimate the broiler’s body weight. The dimensions (minimum and maximum height of standing and sitting birds, head to tail length and width between wings), body volume, and area and animal position were correlated with the measured weight using a multilinear regression algorithm. Data was collected using 80 broilers (Cobb) from 8 days old to 34 days old. Preliminary results indicate that the broiler’s body weight can be estimated from their body dimensions, volumes, and area using a multilinear regression model (R² = 0.96). The results indicate that this model can be used as a tool to effectively and practically estimate the body weight of the broiler during production phase.