JoyCaption Batch Image Captioner

Created an image processing tool focused on batch captioning, since JoyCaption’s demo only captions one image at a time.

Installation

1. Clone this repository

git clone https://github.com/LightCyan01/JoyCaption-Batch-Image-Captioner.git
cd JoyCaption-Batch-Image-Captioner

2. Clone the JoyCaption model

git clone https://huggingface.co/fancyfeast/llama-joycaption-beta-one-hf-llava

3. Create venv and activate

python -m venv venv

venv\Scripts\activate

4. Install requirements

For GPU:

pip install -r requirements-cuda.txt

For CPU:

pip install -r requirements.txt

4. Run GUI

python image_captioner.py

5. CLI Alternative

python batch_caption.py <image_folder> [options]

Examples:

python batch_caption.py "C:\MyImages"

# Overwrite existing captions
python batch_caption.py "C:\MyImages" --overwrite

# Use descriptive style
python batch_caption.py "C:\MyImages" --style descriptive

# Custom model path
python batch_caption.py "C:\MyImages" --model "path/to/custom/model"

Options:

• --style : Caption style (training, descriptive, straightforward) - Default: training
• --overwrite : Overwrite existing caption files
• --model : Custom model path - Default: llama-joycaption-beta-one-hf-llava

Caption Styles:

• training (default): Short, factual descriptions perfect for LoRA training
• descriptive: Long, detailed descriptions
• straightforward: Concise, objective captions

Creates .txt files alongside each image.

Credits:

• JoyCaption: https://github.com/fpgaminer/joycaption