如何使用Python的imageio库不加载图片到内存获取图片尺寸？

Absolutely! You’re spot-on that most common image formats (like JPEG, PNG, BMP) store their dimension data in the file header, so we can retrieve those details without loading the full pixel buffer into memory. And the good news is imageio has built-in ways to do this—no need to switch to other libraries.

Here are the best approaches depending on your imageio version:

For Imageio v3 (Recommended, Latest Version) #

Use immeta() to fetch only the file's metadata, which includes dimensions, without loading any pixel data:

import imageio.v3 as iio

# Fetch metadata from the file header
metadata = iio.immeta(filepath, exclude_applied=False)

# Extract dimensions (keys are consistent for most common formats)
width = metadata["width"]
height = metadata["height"]

This method is lightweight and efficient, as it only parses the necessary header information.

For Imageio v2 (Legacy Version) #

If you're still using the older v2 API, you can use get_reader() and access the metadata without reading the full image:

import imageio

with imageio.get_reader(filepath) as reader:
    meta_data = reader.get_meta_data()
    width = meta_data["width"]
    height = meta_data["height"]

The context manager ensures the reader is properly closed after retrieving the metadata.

Quick Notes: #

• This works for all major image formats where dimensions are stored in the header. For more obscure formats that don't follow this structure, you might still need to load part of the file, but this covers 99% of everyday use cases.
• Compared to your original imageio.imread(filepath).shape approach, these methods use drastically less memory—critical when working with large images like raw photos or high-resolution scans.

内容的提问来源于stack exchange，提问作者Zorglub29