In recent years, 3D face reconstruction has become a research hotspot in computer graphics and computer vision. Most current 3DMM-based methods focus on learning displacement maps to recover high-frequency facial details. However, they focus less on learning mid-frequency facial details and introduce displacement maps with noise, decreasing face reconstruction accuracy. Thus, this work presents a novel approach to regressing accurate and detailed 3D face shapes. First, we design a novel feature consistency loss to recover mid-frequency facial details. Specifically, we exploit the powerful CLIP as prior knowledge of faces to extract geometric and semantic features, which helps guide the reconstructed 3D geometric details to match local details in the input image. Furthermore, we propose a parameter refinement module to learn fine-grained features. It helps to obtain accurate model parameters and improve the accuracy of facial reconstruction. Extensive experiments on a FaceScape and a REALY benchmark demonstrate that our method outperforms several state-of-the-art methods in reconstruction accuracy. Furthermore, comprehensive qualitative results show that our approach achieves better visual performance than existing methods.

This work investigates a multi-product parallel disassembly line balancing problem considering multi-skilled workers. A mathematical model for the parallel disassembly line is established to achieve maximized disassembly profit and minimized workstation cycle time. Based on a product’s AND/OR graph, matrices for task-skill, worker-skill, precedence relationships, and disassembly correlations are developed. A multi-objective discrete chemical reaction optimization algorithm is designed. To enhance solution diversity, improvements are made to four reactions: decomposition, synthesis, intermolecular ineffective collision, and wall invalid collision reaction, completing the evolution of molecular individuals. The established model and improved algorithm are applied to ball pen, flashlight, washing machine, and radio combinations, respectively. Introducing a Collaborative Resource Allocation (CRA) strategy based on a Decomposition-Based Multi-Objective Evolutionary Algorithm, the experimental results are compared with four classical algorithms: MOEA/D, MOEAD-CRA, Non-dominated Sorting Genetic Algorithm Ⅱ (NSGA-Ⅱ), and Non-dominated Sorting Genetic Algorithm Ⅲ (NSGA-Ⅲ). This validates the feasibility and superiority of the proposed algorithm in parallel disassembly production lines.