Wrangle EXIF Data in Python

A Python package for wrangling EXIF data extracted from images using Phil Harvey's EXIFTool.

Set-up

Install pyexifwrangle with pip

$ pip install pyexifwrangle

Install Phil Harvey's EXIFTool

Install Phil Harvey's EXIFTool from https://exiftool.org/. This site has installation instructions if you need them.

Usage

Get EXIF data

After installing EXIFTool, you can extract EXIF data from every image in a folder, including subdirectories, save the results in a csv file, and return the results in a Pandas DataFrame.

The filename_col is the name of the column that contains the image filenames. The function get_exif extracts and saves the EXIF data using ExifTool then runs the function read_exif to load the EXIF data into a Pandas DataFrame. The filename column is checked for files that start with "." and if any such files are found they are deleted from the DataFrame.

import pyexifwrangle.wrangle as wr

df = wr.get_exif(input_dir='path/to/images', output_csv='path/to/output.csv', filename_col='SourceFile')

Load the EXIF data

If you already used get_exif to extract and save the EXIF to a csv file, you can use read_exif to load the csv file into a DataFrame. In this case, the output of get_exif is the same as the output from read_exif.

import pyexifwrangle.wrangle as wr

df = wr.read_exif(path='path/to/output.csv', filename_col='SourceFile', encoding='utf-8')

The function read_exif uses the Pandas package to load the csv file into a DataFrame. The parameter filename_col is the name of the column that contains the filenames of the images. Any filenames that start with "." are deleted from the DataFrame.

The function read_exif uses the pandas.read_csv to load the csv file. In most cases the default encoding 'utf-8' worked on the EXIF data, but occasionally I have needed to change the encoding. The encoding argument is passed onto pandas.read_csv.

Make columns from folder names

I often organize my images into folders and sub-folders, and I usually want to capture the folder names in columns in the DataFrame. The function filename2columns does this for me.

For example one of my projects has the following folder tree:

├── Samsung_phones  # main directory
│   ├── s21  # model
│   │   ├── s21_1  # phone name
│   │   │ 	├── blank  # scene type
│   │   │	│	├── front  # camera
│   │   │	│	│	├──image1.jpg
│   │   │	│	│	├──image2.jpg
│   │   │	│	│	├──...
│   │   │	│	├── telephoto
│   │   │	│	│	├──image1.jpg
│   │   │	│	│	├──image2.jpg
│   │   │	│	│	├──...
│   │   │	│	├── ultra
│   │   │	│	│	├──image1.jpg
│   │   │	│	│	├──image2.jpg
│   │   │	│	│	├──...
│   │   │	│	├── wide
│   │   │	│	│	├──image1.jpg
│   │   │	│	│	├──image2.jpg
│   │   │	│	│	├──...
│   │   │ 	├── natural
│   │   │	│	├── front  
│   │   │	│	├── telephoto
│   │   │	│	├── ultra
│   │   │	│	├── wide
│   │   ├── s21_2 
│   │   │ 	├── blank
│   │   │	│	├── front
│   │   │	│	├── telephoto
│   │   │	│	├── ultra
│   │   │	│	├── wide
│   │   │ 	├── natural
│   │   │	│	├── front  
│   │   │	│	├── telephoto
│   │   │	│	├── ultra
│   │   │	│	├── wide

Extract the folder names from the images' absolute filepaths and make a new column for each folder.

df = wr.filename2columns(df=df, filename_col='SourceFile', columns=['model', 'phone', 'scene_type', 'camera', 'image'])

Search for missing EXIF data

Find images missing a specific EXIF data column. For example, search the DataFrame for images that don't have an Aperture.

missing = wr.check_missing_exif(df=df, column='Aperture')

Count images per group(s)

Group images by column(s) and count the number of images per group.

counts = wr.count_images_by_columns(df=df, columns=['model', 'phone', 'scene_type', 'camera'])

Optionally, you can sort the output of count_images_by_columns

counts_sorted = wr.count_images_by_columns(df=df, columns=['model', 'phone', 'scene_type', 'camera'], sorted=
['phone', 'camera', 'Aperture'])

Wipe GPS Information from the EXIF data

Delete all GPS related tags from an image's EXIF data with wipe_gps. This function removes GPS information from the EXIF data stored in the image file itself. The image itself is not changed; only its EXIF data is changed.

This function DOES NOT delete GPS information from an EXIF csv file that has already been saved with get_exif, nor does it delete GPS information from a DataFrame that is currently loaded in Python. Rerun get_exif to generate an EXIF csv file and DataFrame without GPS data.

wr.wipe_gps(path='path/to/image.jpg')