Open Image Search

What Open Image Search is

Open Image Search is a free, open-source reverse image search prototype for museums and graphical collections working with digitized prints and paintings. Its hosted web application searches more than 20,000 prints from the Graphische Sammlung of ETH Zürich, while organizations can clone the project to run retrieval locally on their own image data. The search process extracts features with VGG16 and uses a k-nearest neighbors model to find similar images. Collections can attach metadata through a CSV file using column names specified in the project documentation. The mobile-friendly interface can send search requests from a phone camera. Behind the interface, a Django REST Framework API reformats uploaded images in memory before sending them to the feature extraction model, then uses the results to retrieve metadata for top matches. The project lists Vue, Django, TensorFlow, scikit-learn, SQLite, and Nginx among its technologies. It is MIT-licensed. The project describes itself as a prototype in development, and its GitHub repository was archived on September 1, 2026, making it read-only. Inquiries go to the ETH Library Lab contact form.