Automated image analysis techniques have revolutionized agricultural practices, offering valuable insights for crop management. In this paper, we present a novel approach for leaf and grape monitoring. It leverages a machine learning-based image segmentation algorithm and temporal indexes computed using vineyard infield images. We extend the scope of an existing dataset containing semantic segmentation masks of grapes, adding support to the leaf detection tasks. Specifically, we implement a data augmentation technique through which we programmatically add synthetic leaves to existing input images. Then, we train a segmentation algorithm using DaFormer with Domain Adaptation via Cross-domain Mixed Sampling (DACS) for accurate segmentation of both leaves and grapes. Additionally, we compute optical indexes designed for grapes and leaves analysis using the corresponding segmented images. We analyze the temporal evolution of these indexes to provide valuable insights into the growth and health status of leaves and grapes. The interpretation of such indexes could offer important information to winegrowers, enabling them to make better-informed decisions regarding crop management. To promote future work on this topic, we release the collected dataset, comprising the original images along with segmentation masks generated by our model.