* add modes and documents for vehicle and pedestrian detection

* remove needless output images

* add class names

* add Configure for training

* fix table in READ.md

* compress the images in the demo/output

* change the score_threshold for vehicle inference

* change Yolo v3 to YOLOv3