videos/ holds test videos
bg_sub.py - simple backgroup subtraction - KNN or MOG2 md.py - motion detector using dilation, background subtraction, and contours to identify 'objects' (like vae-md)
dense_flow.py - dense flow est. Color as direction point_flow.py - point flow est.
object_trackers.py - multiple object detectors, given an initial bounding box dnn_track.py - MobileNet SSD tracker md_tracker.py - complex tracker, using MD to identify potential things to track, then dlib's Mosse pixel correspondance tracker
face_cascade.py - Face detection using Haar face_dnn.py - Face detection using ___ (Resnet10?)
face_extractor.py - utility to extract all faces (at a interval, e.g. every 30 frames) and save them, for building face templates face_embedder.py - utility to generate embeddings of all faces face_train_model.py - train Support Vector Machine using embeddings face_recognizer.py - detect faces in video, run them through face embedder, compare to SVM for recognition
model/ holds models + weights output/ holds pickle'd embeddings / vectors faces/ holds folders per-person, and a number of faces per
GOTURN:
wget http://cs.stanford.edu/people/davheld/public/GOTURN/trained_model/tracker.caffemodel