Document Type
Dataset
Source Publication Title
Simulation-Based Multi-Fillet Evaluation of Woody Breast Poultry Fillets
DOI
https://doi.org/10.32855/dataset.2026.02.049
Production/Collection Date
2025 December
Production/Collection Location
Arlington, TX, USA
Depositor
William J. Beksi
Deposit Date
6-23-2026
Data Type
3D meshes (.obj files)
Abstract
This Synthetic-Chicken-Fillets dataset contains 1,000 synthetic 3D meshes designed to capture the natural variance and size diversity of real broiler fillets. The collection was developed to test automated woody breast detection algorithms within a physics-based simulation environment. We utilized a seed dataset of 2D depth maps derived from 40 real-world RGBD point cloud scans. These real depth maps were fed into a few-shot transfer learning pipeline using a generative adversarial network architecture. The resulting generated depth maps were reconstructed back into 3D meshes. The length and thickness of each mesh were randomly scaled based on physical measurements of real broiler breast fillets. Because the generated meshes lacked a base, their boundaries were stitched to a flat bottom to form closed, watertight objects.
Disciplines
Agriculture | Artificial Intelligence and Robotics
Publication Date
Summer 6-15-2026
Language
English
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Recommended Citation
Sen Mukherjee, Chirantan; Yoon, Seung-Chul; and Beksi, William J., "Synthetic-Chicken-Fillets" (2026). Agriculture. 1.
https://mavmatrix.uta.edu/rvl_agri_datasets2/1
Comments
Project website: https://robotic-vision-lab.github.io/woody-breast-sim