Automated Image Object Detection and Indexing Workflow

This n8n workflow automates the process of analyzing an image to identify, crop, and index objects for enhanced image search capabilities. The workflow begins with a manual trigger, allowing users to start the process manually. It then fetches a source image from a URL, which can be customized within the workflow variables. The image is sent to Cloudflare’s AI service that uses the Detr-Resnet-50 model to classify objects within the image, filtering results with a confidence score of 0.9 or higher. Each identified object is cropped from the original image based on its bounding box coordinates, creating individual object images. These cropped images are uploaded to Cloudinary for storage and further processed to generate URLs. The workflow then indexes the object images along with metadata into Elasticsearch, creating a searchable database of objects within images. Throughout the process, sticky notes provide visual guidance and documentation, making it easier to understand each step and its purpose. This workflow is ideal for digital asset management, creating granular image search, or building an object-based image catalog where users need to quickly identify and retrieve specific objects from images.