Advanced image understanding, as an API
A cloud service that looks at your images and returns rich, per-image intelligence in seconds — the building blocks for any product that needs to understand photos. Powered by vision-language models and image embeddings, it goes beyond labels to read aesthetics, people, framing, and how images relate. Your app sends photos; the API streams back structured results as each one finishes.
Early access — available to design partners
Understanding photos by hand doesn’t scale
Whether you’re building a photobook, curating a gallery, moderating uploads, or organizing a media library, the hard part is the same: judging what each image shows and how good it is. Doing that by hand across hundreds or thousands of images is slow, inconsistent, and doesn’t scale.
Every photo, understood
Aesthetic quality
A score for how visually strong each photo is — so the best shots surface first.
Faces
Where people are in the frame — the anchors of any personal photobook.
Region of interest
The part of the image that matters most — for smart, subject-aware cropping.
Framing / zoom
Close-up, mid, or wide — so layouts mix shot types naturally.
Visual similarity
A fingerprint of each photo — to group related shots and skip near-duplicates.
Made to ship
Streamed results
Submit a batch and get an immediate acknowledgment; results stream back over a real-time channel as each photo is analyzed, so you never poll for status.
Parallel processing
Photos in a batch are processed in parallel, so a 10-photo batch and a 500-photo batch return on a similar timeline.
Fully managed cloud service
The analysis runs on our infrastructure — there are no models to host, no GPUs to provision, and nothing to keep updated on your side.
Simple to integrate
One REST call to submit, one real-time subscription to receive. No ML expertise required on your side.
Vision-language models + embeddings
Vision-language models read image content and quality; image embeddings turn every photo into a vector for similarity and near-duplicate detection.
Built for a real workload
The analysis pipeline was first built to power automatic photobook design, so it is shaped around real, unsorted camera rolls rather than clean benchmark sets.
The perception layer for your product
The API’s structured, per-image output plugs straight into whatever you’re building — auto-layout and smart cropping for a photobook, ranking and filtering for a gallery, safety checks for user uploads, or organization for a media library. You get clean signals; your product decides what to do with them.
Where your images go
Processed, then discarded. You send images to the cloud API; they are analyzed, the results are returned, and the images are not retained after analysis.
Hosted in Switzerland and the EU. The cloud service runs in Swiss and EU regions.
Building on it and need the specifics for a security review? Ask us about data handling.
One API. Every photo, understood.
Image Analysis is in early access. Tell us what you’re building and we’ll set you up with a design-partner integration.
info@edeltech.ch