Search

Semi-supervised instance segmentation algorithms are mainly divided into algorithms based on pseudo-label generation and algorithms based on transfer learning. The algorithms based on pseudo-label generation need to design a specific pseudo-label generation process, but the process is not scalable for different types of source tasks. The algorithms based on transfer learning that started late have relatively high scalability, but the algorithm research ideas are relatively simple. To expand the research on semi-supervised instance segmentation based on transfer learning, this paper proposes a feature transfer-based semi-supervised instance segmentation algorithm Feature Transfer Mask R-CNN (FT-Mask). The FT-Mask algorithm is more scalable than algorithms based on pseudo-label generation and can be used to transfer knowledge from different types of source tasks. Compared with other semi-supervised instance segmentation algorithms based on transfer learning, FT-Mask uses the feature transfer method to achieve semi-supervised instance segmentation for the first time. The experimental results show that the FT-Mask model improves the semi-supervised instance segmentation accuracy of the Mask R-CNN benchmark model through the semi-supervised learning process, and can achieve effective transfer learning. [ABSTRACT FROM AUTHOR]

Accurately and rapidly predicting the shrink mold shape of direct injection-expanded foam molding is an important and difficult task. This molding method is widely used by sports shoe sole manufacturers to create shock-resistant materials. Modifying the shrink mold shape using the numerical optimization method is crucial to rapidly obtaining the correct shrink mold size. This study uses a series of rectangle specimens to identify the relationship between the thermal heating of molding and the expansion ratio of ethylene vinyl acetate foam material and then uses this thermal expansion ratio to simulate expansion behavior. The experiments in this study also use the actual shoe sole type, which has the original three-dimensional (3D) shape, and use the proposed simulation method to obtain the simulation expansion shape. This study also develops an optimization algorithm based on 3D registration and the Newton–Raphson method to obtain the shrink mold shape. We also manufactured the shrink mold and obtained the shoe sole product to compare any discrepancies between the product and the original 3D shape. The results of this method meet the requirements of the shoe sole factory (i.e., achieve a difference of less than 3 mm). [ABSTRACT FROM AUTHOR]