Recommends suitable selfie poses based on the party size and gives a score based on pose accuracy.
Demo video on Youtube
json file names should be followed by the size of party ex) pose_name_4, pose_name_2, ...
REFERENCE
for detecting multiple people: yolov3
to install darknet to be able to use yolo models,
useful Youtube tutorial for extracting landmark of each person detected: mediapipe holistic model