3D Reconstruction of Live Chickens in Poultry Houses
Muni, Aneri Dhirendra
MetadataShow full item record
Poultry houses require daily monitoring to ensure animal health and proper house operation. One task involves observing the average growth rate of the house to adjust the daily feed. In addition to being labor intensive and time consuming, it is difficult for the farm owners to find consistent labor to fill these jobs. This project looks at the possibility of estimating the weight of a chicken based on the volume of the chicken as captured by a 3D model. We present a system capable of reconstructing dynamic scenes, i.e. chickens in a poultry house, by fusing together depth scans captured using a Microsoft Kinect. Like DynamicFusion, our approach involves discretizing the live depth frame into nodes and estimating individual 6D transformations before fusing them together to reconstruct scene geometry. This approach doesn’t use any prior model of template, making it applicable to a wide range of dynamic objects and scenes.