Crop photos and extract text with AI
Resize or crop a picture, examine its metadata and extract visible text using a local image-processing workflow.
You can crop photos and extract text with AI when a product photo or scanned label needs a few ordinary corrections. The image skill connects an agent to local tools for resizing, format conversion, metadata inspection and optical character recognition.
For a small catalogue, keep the original photograph and work on a copy. Reading the label before making the web copy smaller leaves clearer lettering for OCR. A final metadata inspection can catch location information that should not travel with the published picture.
What it does
- Crop a photograph to the part that matters.
- Resize, rotate or flip a copy for a particular layout.
- Read embedded metadata and strip it from a separate output.
- Extract visible text with an appropriate OCR language.
- Produce a compressed JPEG or choose a different format explicitly.
Install and use
- Install the skill, Python and uv, plus the local tools you need, such as ImageMagick, ExifTool and Tesseract. Use an agent authorised to see the source.
- Ask “Read this package label, crop the photograph and save a web-sized copy. Keep the original.”
- Compare the result at its actual display size and verify names and numbers against the picture.
FAQ
Source
Made by timelapsetech. Original skill: image, published on GitHub under the MIT licence.
View on GitHub