Revolutionizing Grape Cultivation: AS-SwinT and the Future of Automated Berry Thinning
Berry thinning in grape cultivation is a labor-intensive process carried out by skilled workers, but the scarcity of laborers due to an aging population has led researchers to develop an intelligent machine vision system for automated berry thinning. Deep learning and image processing techniques are used to count and locate berries, but challenges remain in accurately detecting small berries due to their size, color similarity to leaves, and dense packing. In this study, researchers used a machine vision system called AS-SwinT to enhance the process of grape berry counting. The system demonstrated superior performance in detecting small berries and outperformed other models in artificial and natural environments. The study also explored the number estimation of berries using a linear regression model. Challenges for future research include the non-detection of heavily shaded or tiny berries, as well as the impact of environmental conditions on detection accuracy. Overall, this research presents a significant advancement in automated grape berry counting and has the potential to revolutionize table grape cultivation.