Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/QwenLM/Qwen3-VL/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

The qwen-vl-utils package provides helper functions for processing and integrating visual language information with Qwen-VL Series Models. It handles image and video loading, resizing, and formatting for use with Qwen2VL, Qwen2.5VL, and Qwen3VL models.

​
Installation

pip install qwen-vl-utils

​
When to Use

Use qwen-vl-utils when you need to:
  • Process images from various sources (local files, URLs, base64, PIL.Image objects)
  • Extract frames from videos for vision-language tasks
  • Automatically resize images and videos to optimal dimensions
  • Prepare vision inputs for Qwen-VL model processors

​
Key Functions

process_vision_info

Main function to extract and process all vision information from conversations

fetch_image

Load and resize images from files, URLs, or base64 strings

fetch_video

Extract and process video frames with configurable parameters

smart_resize

Intelligently resize images while maintaining aspect ratio

​
Quick Example

from transformers import Qwen2VLForConditionalGeneration, AutoProcessor
from qwen_vl_utils import process_vision_info

messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "image": "file:///path/to/image.jpg"},
            {"type": "text", "text": "Describe this image."}
        ]
    }
]

processor = AutoProcessor.from_pretrained(model_path)
model = Qwen2VLForConditionalGeneration.from_pretrained(
    model_path, torch_dtype="auto", device_map="auto"
)

text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
images, videos = process_vision_info(messages)
inputs = processor(text=text, images=images, videos=videos, padding=True, return_tensors="pt")

generated_ids = model.generate(**inputs)

​
Supported Input Formats

​
Images

  • Local file paths: file:///path/to/image.jpg
  • HTTP/HTTPS URLs: http://example.com/image.jpg
  • Base64 encoded: data:image;base64,/9j/...
  • PIL.Image objects: Direct PIL.Image.Image instances

​
Videos

  • Local video files: file:///path/to/video.mp4
  • HTTP/HTTPS URLs: http://example.com/video.mp4
  • Frame sequences: List of image paths representing video frames

Build docs developers (and LLMs) love

Get started for freeTalk to us