synthetic 3D fillet dataset
收藏资源简介:
该合成三维鸡胸肉片数据集由德克萨斯大学阿灵顿分校与USDA研究团队联合创建,旨在通过高保真模拟解决木质化鸡胸肉在线检测难题。数据集包含1000个合成三维网格模型,通过基于StyleGAN2的生成架构结合平滑相似性正则化技术,从40个真实扫描样本中扩展生成,涵盖了家禽肉片形态的自然变异范围。其创建过程包括将点云转化为深度图、利用生成对抗网络合成多样化样本,并重建为可用于物理仿真的三维网格。该数据集主要应用于计算机视觉与农业工程领域,为开发基于顶置摄像头的多肉片同时检测算法提供训练与验证基础,以提升家禽加工线的检测效率与经济效益。
This synthetic 3D chicken breast slice dataset was co-developed by the University of Texas at Arlington and the USDA research team, aiming to solve the on-line detection problem of wooden breast chicken meat via high-fidelity simulation. The dataset contains 1000 synthetic 3D mesh models, which are expanded from 40 real scanned samples using a StyleGAN2-based generative architecture combined with smooth similarity regularization technology, covering the full range of natural morphological variations of poultry slices. Its creation process includes converting point clouds into depth maps, synthesizing diverse samples through generative adversarial networks (GANs), and reconstructing them into physically simulatable 3D meshes. This dataset is mainly applied in the fields of computer vision and agricultural engineering, providing training and validation foundations for developing simultaneous detection algorithms for multiple slices based on overhead cameras, so as to improve the detection efficiency and economic benefits of poultry processing lines.

- 1Simulation-Based Multi-Fillet Evaluation of Woody Breast Poultry Fillets德克萨斯大学阿灵顿分校·计算机科学与工程系; 美国农业部农业研究服务局·国家家禽研究中心·质量与安全评估研究部 · 2026年




