See everything in your image, not just what you're looking for

It means our AI looks at your image and identifies what's in it — people, objects, scenes, text, brands, colors, lighting conditions. It's like having someone describe the photo to you in detail, except it's a computer model running six different detection passes at once. Don't expect it to read emotions or understand context — it identifies, it doesn't interpret.

Content Analyzer gives you structured data — a list of objects, labels, and confidence scores. Image Describer generates natural-language sentences and paragraphs. Use Content Analyzer when you need raw data (for tagging, filtering, or feeding into another system). Use Describer when you need human-readable text (for alt text, captions, or descriptions).

Accuracy varies with the subject, image quality, lighting, and editing. Confidence scores describe the model's certainty, not a verified probability of correctness, so review important results manually.

No. We detect whether faces are present, estimate basic demographics (age range, apparent gender), and count how many people are in the shot — but we do not identify specific individuals or match faces against any database. We made a deliberate choice not to go there on privacy grounds.

English, Chinese, Japanese, Korean, Arabic, Cyrillic scripts, and most Latin-based languages — basically anything you'd commonly see on signs, screens, or documents. Handwriting is hit or miss — clear printed text works great, messy cursive, not so much.

The web interface currently handles one image at a time. A public batch API is not available yet; you can register your interest on the API Access page.