Integrating deep learning image analysis into Web GIS applications: A Hybrid Flask - Spring Boot architecture for automated place detection

This paper presents a hybrid web GIS architecture that integrates deep learning–based image recognition with a robust spatial data management interface. Developed using a combination of Spring Boot (Java), Leaflet (JavaScript), and PostgreSQL/PostGIS, the application provides users with a rich interface for route planning, layer visualization, and CRUD operations on categorized places and groups. At the core of this study is the novel integration of a lightweight Python Flask microservice, which leverages OpenAI’s CLIP model and EXIF metadata extraction to automate the classification and geolocation of uploaded images. This hybrid system allows users to add new places either manually by entering coordinates, selecting categories, and uploading an image or automatically, using Smart Detection mode. In this second mode, the image becomes the primary input source, from which the application semantically infers the appropriate group (e.g., museum, park, church) and extracts GPS coordinates from the image’s EXIF data. This dual-mode input architecture enhances both user flexibility and data accuracy while demonstrating a practical fusion of deep learning and GIS through modern web frameworks.