I keep a library of roughly a thousand photos—mostly different remote controls—that only fifteen teammates and I need to access. What I’m missing is a quick, Google-Lens-style search tool: take one photo, drop it into the site, and instantly see the five most similar images already stored. The site must stay completely private, so user management for exactly 15 accounts (secure login, password reset) is essential. On the technical side the AI has to recognise shapes and outlines, colours and patterns, plus any numbers or letters that appear, then rank the closest matches. Optional tag or category filters would be useful so we can narrow a search when the library grows. I’m flexible on the stack—Python/Django + TensorFlow or PyTorch, Node with OpenCV, even a self-hosted cloud-vision wrapper—as long as the solution runs on my own server and can scale as more images are added. Deliverables • Responsive web interface with upload and search workflow • Computer-vision back-end returning the top five matches per query • Admin panel to bulk-import photos, manage tags/categories, and handle the fifteen user accounts • Clear deployment script or Docker setup so I can install it on an Ubuntu VPS Acceptance test: I upload a fresh photo of a remote control and, in under two seconds, the site displays the five closest images with confidence scores. Please outline the libraries or models you’d use and link to any similar work you’ve completed.