Image & Text Assistant

Describe an image or read a cropped line of printed text. AI runs in your browser.

First use downloads model files and may take several minutes. Use a desktop browser and keep this tab open. Check every result against the image.

1. Choose your image and task

Image (PNG, JPEG, or WebP; up to 10 MB) Task Describe image Read printed text Run both on this image

For text recognition, crop closely around one printed text line. Full documents and ordinary photos without text are outside this OCR model's scope.

Correct printed text (optional, for evaluation) Analyze image Clear session

2. Review the results

Generated image description Recognized printed text

No model has run yet.

Download session results (CSV)

No account, API key, or paid inference service is needed to use this app. Selected images are processed locally by this application. Model files come from Hugging Face; the library comes from a CDN. Avoid confidential images.

Models: Xenova/vit-gpt2-image-captioning and Xenova/trocr-small-printed. Clearing removes session inputs and results; downloaded model files may remain cached.