Improved Image Acquisition for Accurate Pruning Wood Weight Estimation via Image Analysis in

Pruning wood mass is a key factor in grapevine management, influencing vine growth and yield. Accurate assessment of pruning wood weight helps viticulturists regulate vine balance, optimize shoot production, and improve fruit quality. However, traditional manual methods are time-consuming and labour-intensive, creating a need for more efficient, non-destructive alternatives. While digital imaging technology has advanced in viticulture, enabling non-invasive methods for yield estimation and canopy assessment, research on pruning wood weight estimation is limited due to the need to use artificial backgrounds for background homogenisation. This study evaluates image analysis for estimating pruning wood weight, focusing on the impact of image acquisition conditions rather than segmentation techniques. The goal of this research was to address challenges related to the use of artificial backgrounds during image acquisition examining the relationship between pixel counts in images of pruning wood and actual wood weight under two acquisition modes. Results showed that using an artificial background did not significantly improve estimation accuracy (R2=0.70) compared with no background acquisition mode (R2=0.69). This suggests that a simple camera system could be sufficient for effective pruning wood assessment. Moreover, the possibility to acquire images without the need for an artificial background could enable the use of cameras mounted on vehicles performing routine vineyard operations. This would allow for the collection of data on wood weight from different vineyard areas, providing valuable insights into vine vigour.