page	title	tags	date
https://idle.run/face-progress	Face Alignment for Growth Progress	python opencv	2019-05-20

Uses OpenCV to align the face in a series of photos with optional manual fine tuning.

Prerequisites

DLib Feature Model

Download the 68 feature model for face shape prediction (into the repository alongside progress.py)

wget https://github.com/davisking/dlib-models/raw/master/shape_predictor_68_face_landmarks.dat.bz2
bunzip2 shape_predictor_68_face_landmarks.dat.bz2

Environment Setup

• Create a directory input filled with images to be processed
• Create a directory output for the result images to be written
• Choose a single reference photo to use and put it in the input directory
• Run ./progress.py
• Adjust the image as desired using Controls described below and save
• Copy the resulting image from output and overwrite reference.jpg with the new image

Usage

• Load up the input directory with all images and run ./progress.py to process them
• Images will appear one-by-one with auto rotation based on detected features.
• Use the Controls described below to fine tune image alignment as closely as needed
• Using fix-photo-dates on the output may be helpful

Controls

• q and e rotate left and right
• wsad pan image
• t and g zoom in and out (Ideally all images should be taken from the same distance so that zoom is not required)
• r overlay reference image with working image for alignment assist
• <enter> save image and open next
• <spacebar> skip image and open next

Syncing from Android

Here is a simple script to sync new photos only from an Android device over adb

In this example I'm using the Open Camera app set to save images to the folder /sdcard/Progress - adjust as required.

cd input
adb ls /sdcard/Progress | grep -e "\.jpg$" | cut -d" " -f4 | grep -v -f <(echo __; ls -1; ls -1 ../done) | xargs -I{} -n 1 adb pull /sdcard/Progress/{}

Convert to Video

Here is an example of converting the images to a video. Adjust the cropping params as required.

ffmpeg -r 10 -f image2 -pattern_type glob \
  -i 'output/*.jpg' -filter:v 'crop=in_w*6/12:in_h*7/12:in_w*3/12:in_h*1/12' \
  -an -c:v libx264 -pix_fmt yuv420p -crf 25 \
  vid.mp4