Technology

Technical Summary

• Objective: Develop an image similarity search application to enhance access to the Cline Library's Special Collections and Archives (SCA).
• Front-End: Utilize React for a dynamic and responsive user interface.
• Back-End: Employ Node.js and Express for API handling and server-side logic.
• Hosting: Use AWS for scalable and flexible cloud-based infrastructure.
• Machine Learning Model: Vision Transformer (ViT) for embedding and image similarity detection.
• Tools and Libraries:
  • TensorFlow for embedding generation.
  • Milvus for efficient vector storage and querying.
• Key Challenges:
  • Ensuring seamless integration between front-end and back-end.
  • Handling large-scale vector databases efficiently.
  • Maintaining scalability for future growth.
  • Providing a user-friendly interface for seamless interaction.
• Outcome: A responsive application offering precise and efficient search functionalities for accessing the SCA image database.