Tags recognized faces in images in Adobe Bridge compatible format.
Currently only tested and working on Windows 10.
Usage:
gesichtserkennung.py [-v] --references=<PATH> [--tolerance=<VALUE>] <PATH>...
-r PATH, --references <PATH> Path to the folder with reference images
-t VALUE, --tolerance <VALUE> Tolerance for face detection (higher: more false positives, less false negatives) [default: 0.55].
-v, --verbose Verbose output
Example:
C:\>exiftool photo_2022-05-22_05-24-38.jpg
ExifTool Version Number : 12.44
File Name : photo_2022-05-22_05-24-38.jpg
Directory : c:/
File Size : 133 kB
File Modification Date/Time : 2022:08:11 00:07:25+02:00
File Access Date/Time : 2022:08:11 00:07:25+02:00
File Creation Date/Time : 2022:08:10 23:57:59+02:00
File Permissions : -rw-rw-rw-
File Type : JPEG
File Type Extension : jpg
MIME Type : image/jpeg
Image Width : 640
Image Height : 640
Encoding Process : Baseline DCT, Huffman coding
Bits Per Sample : 8
Color Components : 3
Y Cb Cr Sub Sampling : YCbCr4:2:0 (2 2)
Image Size : 640x640
Megapixels : 0.410
C:\>face_tagger -r references photo_2022-05-22_05-24-38.jpg
Starting analysis of photo_2022-05-22_05-24-38.jpg ...
Analysis of photo_2022-05-22_05-24-38.jpg finished. Proceeding adding recognized persons to the metadata.
Found <REDACTED> in this image. Updating metadata!
<REDACTED2> is already tagged in this image. Not adding them again.
3 unknown person(s) detected. Please check manually!
C:\>exiftool photo_2022-05-22_05-24-38.jpg
ExifTool Version Number : 12.44
File Name : photo_2022-05-22_05-24-38.jpg
Directory : c:/
File Size : 137 kB
File Modification Date/Time : 2022:08:11 00:05:46+02:00
File Access Date/Time : 2022:08:11 00:05:46+02:00
File Creation Date/Time : 2022:08:10 23:57:59+02:00
File Permissions : -rw-rw-rw-
File Type : JPEG
File Type Extension : jpg
MIME Type : image/jpeg
Current IPTC Digest : c934de6c2ab7fa777532fc8a9307233a
Keywords : <REDACTED>, <REDACTED2>
Application Record Version : 4
XMP Toolkit : Image::ExifTool 12.44
Subject : <REDACTED>, <REDACTED2>
Hierarchical Subject : <REDACTED>, <REDACTED2>
Image Width : 640
Image Height : 640
Encoding Process : Baseline DCT, Huffman coding
Bits Per Sample : 8
Color Components : 3
Y Cb Cr Sub Sampling : YCbCr4:2:0 (2 2)
Image Size : 640x640
Megapixels : 0.410
Warning, this can take a while depending on the performance of your system. This script automatically uses all the available CPUs to speed it up as much as possible, but it still takes ~ 30 seconds on my computer to process a single image.
The dlib dependency needs CMake to install.
Furthermore exiftool needs to be available in the PATH at runtime.
For the development version:
git clone https://github.com/jonasw234/face_tagger
cd face_tagger
python3 setup.py install
pip3 install -r dev-requirements.txt
For normal usage do the same but don’t include the last line or use pipx and install with
pipx install git+https://github.com/jonasw234/face_tagger
- Small bugfix in debug logging, code style improvements (breaking up long lines)
- Now processes video files (mp4, avi, mkv, mov) with
cv2(analyzes the first video frame only)