Register for free to
view the full innovation profile.
In this invention, a proposed AI-based approach that targets reconstruction of the volume of a broiler chicken from an RGB camera is used.
About
In this invention, a proposed AI-based approach that targets reconstruction of the volume of a broiler chicken from an RGB camera is used. Our proposed system contains eight steps as follows. First: Data acquisition and annotation. Second: Detection of an individual chicken from the scene captured by the camera. Third: Accurately segmenting the chicken's boundary. Fourth: Estimation of the depth information from an RGB image. Fifth: Detection of the chickens landmarking points and pose estimation. Sixth: 3D reconstruction from multiple views. Seventh: Fitting the model which has been trained on optimizing multiple views of a chicken. Eighth: Calculations result in volume-weight regression network.
Key Benefits
Broiler houses that have our technology will have accurate body weight calculations for all birds, not a subset sample or predicted value. This will result in accurate weights to assess flock weight gain, fitness, uniformity, health, and processing plant yields as well as product volume by predicted slaughter day of age.
Applications
Commercial broiler house: Broiler growth prediction Daily broiler body weight Daily broiler body weight variation Daily broiler feed conversion status Daily broiler slaughter parts determination Daily broiler disease/stress predictions Real-time video viewing/data analyses for technical staff/veterinarians to make quick solution decisions involving feed, feed restriction, water supplements, and health care Commercial hen/pullet house: Pullet growth prediction Daily pullet body weight Daily pullet body weight variation Daily pullet feed conversion status Daily pullet disease/stress predictions Real-time video viewing/data analyses for technical staff/veterinarians to make quick solution decisions involving feed, feed restriction, water supplements, and health care