Multimodal RGB-D Image Dataset for the Estimation of Individual Body Weights and Carcass Measurements in Turkeys

This paper describes a multimodal dataset collected to support research on non-invasive body weight monitoring in poultry. The dataset comprises spatio-temporally aligned RGB and depth images acquired from 30 individually identified male turkeys monitored from 37 to 133 days of age using a ceiling-mounted camera system. Ground truth body weight measurements were obtained through regular manual weighing, and carcass processing data were recorded at the end of the production cycle. Individual birds were tracked over time using color-coded wing tags, enabling longitudinal analysis at the animal level. The dataset includes paired image data, segmentation masks, individual identifiers, and corresponding body weight and carcass measurements. Data were collected under controlled experimental conditions and organized to facilitate linkage between image data and ground truth records. This resource can support the development and evaluation of computer vision methods for body weight estimation, animal detection and segmentation, individual identification, and longitudinal modeling in poultry production systems.