Cocoon shell bionic one-piece vamp virtual sample generation system
By constructing a virtual sample generation system for cocoon-shaped bionic one-piece shoe uppers, and utilizing a multi-dimensional dataset and a multi-scale attention-enhanced U-Net architecture, the system achieves precise mapping between the process parameters and performance requirements of cocoon-shaped bionic one-piece shoe uppers. This solves the problems of long development cycles and high costs in the research and development of cocoon-shaped bionic shoe uppers, enabling rapid and customized design and evaluation, and improving the consistency and comfort of the shoe uppers.
Patent Information
- Application Number
- CN202610578299.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-29
- Publication Date
- 2026-05-29
- Estimated Expiration
- 2046-04-29
AI Technical Summary
The development of cocoon-shaped bionic one-piece shoe uppers is hampered by long development cycles, high material and labor costs, difficulty in quickly balancing support and breathability, large discrepancies between the effects of traditional virtual sample generation and actual molding, lack of a dual-dimensional evaluation system for visual and technological performance, and inability to form a digital closed loop of design, generation, evaluation, and iteration.
A cocoon-shell biomimetic one-piece shoe upper virtual sample generation system was constructed. Through multi-dimensional dataset collection and enhancement, a multi-scale attention-enhanced U-Net architecture model was trained. Combined with hierarchical low-rank adaptive fine-tuning technology, a dual-dimensional evaluation system of visual and process performance was established to achieve accurate mapping and dynamic optimization of process parameters and performance requirements.
It significantly shortens the R&D cycle, reduces costs, achieves integrated molding, improves wearing comfort and durability, quickly adapts to the needs of multiple scenarios, generates virtual samples with high consistency with actual molding, and supports digital closed-loop iterative optimization.