An Object Detection and Segmentation Model-Based Shape Change Estimation Method for Wood Specimen

As an important biomass material, wood is still widely used in modern society. Wood specimens are essential scientific references for classifying, comparing, and identifying wood. The mechanical properties of wood refer to its behavior under external forces, particularly in terms of deformation and failure, including elasticity, plasticity, compressive strength, etc. The mechanical data of wood specimens are crucial reference indices for wood identification. This paper proposes an object detection and instance model-based shape change estimation (ODSE) method for wood specimen, analyzing and measuring area, height and width of the wood specimen by computer vision. Based on the ODSE method, an analysis tool is designed and developed to identify and count the pixels of wood specimens in the provided image data. It achieves measurement and display of those three deformation parameters of wood specimens with pressure changing. Additionally, the recognition accuracy of different weight models obtained through different dataset divisions is explored.