Bisporus Mushroom 3D Reconstruction for Digital Twin of Smart Factory

In recent years, digital twin technology has been gradually applied in the field of agriculture. As a key technology for digital twins in agriculture, 3D reconstruction technology has attracted more and more researchers’ attention. 3D reconstruction of agriculture is difficult because crops are alive, especially for bisporus mushrooms. Bisporus mushrooms are generally grown intensively in narrow spaces in factories, which poses a huge challenge to the layout of the equipment for 3D reconstruction. But there are still some ways to solve the problem of 3D reconstruction of bisporus mushrooms in the factory, except for a small number of extruded mushrooms, most of the bisporus mushrooms can be replaced by regularly shaped geometry, the mushroom cap can be well adapted to the spatial ellipsoid and the space cylinder can express the characteristics of the mushroom stem well. This paper proposes a new method to reconstruction the bisporus mushrooms based on the YOLO v5s network with CBAM attention module. YOLO v5s network with CBAM attention module is used to generate coordinates of bounding boxes to segment the single bisporus mushroom point cloud from the whole point cloud obtained by depth camera, then the point cloud filter denoising algorithm is involved to remove noise spots and the ellipsoid fitting algorithm is used to reconstruct the model of bisporus mushroom caps. The results show that the proposed method can adapt well to bisporus mushroom plant environment in the factory, achieve high-precision fitting, and can be applied to the digital twin technology of bisporus mushroom smart factory in the future.