Grape farming entails a labor-intensive cultivation process, requiring meticulous attention to detail. Once harvested, the grapes undergo a grading process to assess their quality before shipment to customers. For a typical scene of conventional grape grading, the farmer will hold a grape bunch and evaluate its characteristics such as size, color, and density. Subsequently, a grade is assigned to the bunch based on these characteristics.
This grading process is vital for ensuring the quality and market readiness of grapes, yet it presents significant challenges for grape farmers. It heavily relies on the expertise of experienced farmers who can evaluate factors like the size, color, and density of grape bunches. Acquiring this expertise is particularly challenging for new farmers, given the short duration of the grape season. Even experienced farmers may face inconsistencies in grading results due to the subjective nature of the process.
In response to these challenges, a comprehensive grape grading system incorporating Artificial Intelligence, Computer Vision, and Internet of Things technologies was developed. As depicted in the image below, the process begins by receiving four images from which the grape bunches are isolated using an instance segmentation model. These segmented images, along with weight data for each bunch, are processed by a CNN-based model to classify the grape quality. The grading results are then displayed to the user on the client side and simultaneously stored in the data management system for record-keeping.
