PedestrianDetection

Designed by Huanqi Yang in UESTC, 5/2018

Introduction

In this project, I use two ways to realize pedestrian dectection in a video provided by professor Lian Yan, UESTC.

1.transfer video into split pictures, and recognize and mark pedestrian on the pictures, then joint all the pictures.

2.read video frame by frame, and show direct the marked video frame by frame.

1.details about the first way

1)you can use file:[extract pic.py] to tranfer the video into pictures frame by frame.

2)use file[strengthenpic.py] to process the pictures for better results.(optional)

3)use file[detcet.py] to recognize pedestrain in all the pictures.

4)use file[trans to vid] to joint all the recognized pictures into a video.

2.details about the second way

run file[recg_realtime.py], which can show the result video directly.

3.intro:file:[split_images]

the after tranfering images of the video:./video/person.mkv, which was shot by a camera in a driving car.

examples:

4.intro:file:[reg_split_images]

the recognized images

examples: