autocrop

Why we included this project

Anyone who has resized a hundred headshots by hand will appreciate what autocrop does: it finds the largest face in each picture, centers the crop on it, and resizes the result to a consistent width and height. Detection runs on OpenCV's YuNet neural network, and the whole tool installs with a single pip command. You can use it as a command-line filter, feeding image bytes on stdin and getting cropped bytes back on stdout, or import the Cropper class and pass it file paths or OpenCV-style arrays from your own Python code. Because each invocation handles one image, it slots neatly into find, xargs, or GNU parallel for batch jobs, and it accepts a broad range of input formats. For teams that want predictable, scriptable face cropping without maintaining their own detection pipeline, it is a practical drop-in.