Introduction
Pruning weight is a critical metric in viticulture, as it provides an effective means to assess vine vigor, balance, and the potential for crop production in the subsequent year. Traditional approaches to estimating pruning weight are costly, time-consuming, and often require technical expertise or specialized equipment. These limitations make such approaches impractical for small- to medium-sized vineyards. The goal of this study is to create an affordable and easy-to-use system that can be applied broadly by growers or integrated into robotics platforms to support precision viticulture.
Methodology
The authors developed a computer vision-based technique using a standard smartphone camera combined with structured light. This system captures images of dormant vines and processes them to estimate pruning weight. The approach is designed for ease of deployment in both commercial vineyards and potential robotic systems. The system was evaluated on vertical shoot position (VSP) vines and tested in preliminary stages on high cordon procumbent (HC) vines such as Concord grapes.
Results
For VSP vines, the system achieved an $R^2$ value of 0.80, outperforming state-of-the-art computer vision-based methods. For HC vines, the method achieved an $R^2$ value of 0.29, marking the first attempt to use computer vision in this challenging vine architecture. These results indicate that the system is reliable for VSP-trained vineyards and promising for future work on more complex vine types.
Discussion
The simplicity and affordability of the proposed system highlight its potential for immediate adoption by growers. Unlike LiDAR, drones, or multispectral sensors, the setup requires only a smartphone and structured light, making it accessible to vineyards of all sizes. Moreover, the method demonstrates flexibility for both farmer-operated use and robotic integration, supporting broader goals of precision viticulture.
Conclusion
This research introduces a practical, low-cost, computer vision-based method for pruning weight estimation in vineyards. The system demonstrates strong performance for VSP vines and sets the foundation for future work in HC vines. Its affordability and adaptability make it a valuable tool for viticulturists aiming to improve vine balance, optimize crop production, and integrate modern sensing into vineyard management.