This n8n workflow automates the classification of land-use images by leveraging a K-Nearest Neighbors (KNN) algorithm integrated with Qdrant vector search engine. The process begins with receiving an image URL through a workflow trigger. The image is then sent to Voyage AI’s Multimodal Embeddings API, which generates an embedding vector representing the image’s features. This embedding is used to query Qdrant, which returns the closest matching images along with their pre-labeled classes. A majority voting system determines the most common class among these neighbors. If a tie occurs between the top classes, the workflow dynamically increases the number of neighbors queried and repeats the process, ensuring a more accurate classification. The final class label is then returned, providing a reliable land-use categorization. This workflow is highly useful in remote sensing, land management, or environmental monitoring scenarios where automatic land classification from satellite images is needed.