Skip to main content

Image Captioning Example

Overview​

X-AnyLabeling supports image-captioning workflows through its Chatbot and visual question answering tools. Both tools can send the current image to a vision-language model and store or export the generated description for review.

Workflow​

  1. Configure a vision-capable model in the Chatbot.
  2. Load an image directory in X-AnyLabeling.
  3. Use @image with a prompt such as Describe this image in one concise sentence.
  4. Review the generated caption before saving or exporting it.

For structured datasets with configurable fields, use the VQA tool. For batch conversations in ShareGPT format, use the Chatbot's batch-processing and export features.

Always review generated captions before using them as training data, especially descriptions of small objects, text, spatial relationships, and domain-specific terminology.