Module vision 

Vision model captioning for images.

This module provides infrastructure for generating captions from images using vision models. With the vision feature enabled, a BLIP-based captioner is available. Otherwise, only a placeholder implementation exists.

Structs

BlipCaptioner

BLIP-based image captioner using Candle.

CaptionConfig

Configuration for vision captioning.

PlaceholderCaptioner

Placeholder vision captioner that returns no captions.

Enums

CaptionError

Error type for vision captioning operations.

Constants

BLIP_IMAGE_SIZE 🔒

Image size for BLIP preprocessing.

BLIP_MODEL_ID 🔒

BLIP model identifier on HuggingFace Hub.

Traits

ImageCaptioner

Trait for vision-based image captioning.