Skip to content

Latest commit

 

History

History
43 lines (34 loc) · 2.25 KB

README.md

File metadata and controls

43 lines (34 loc) · 2.25 KB
Raw

Computer-Center-People-Counter

Application to count number of people entering and exiting a computer centre in NUS.

Contents

Introduction

This application is used to analyze the usage of AS8 computer centre in NUS. It takes in a CCTV footage and outputs a CSV as well as graph indicating the number of people in the computer centre at each point in time.

Usage

  1. Either get ready some videos or your web cam.
  2. Try the program out with the following command:
  • using web cam:
python people_counter.py --prototxt mobilenet_ssd/MobileNetSSD_deploy.prototxt --model mobilenet_ssd/MobileNetSSD_deploy.caffemodel --output output/<video-filename>.avi --confidence 0.4 --skip-frames <frames-to-skip> --output-csv output/<video-filename>.csv --output-plots <video-filename>
  • using video input:
python people_counter.py --prototxt mobilenet_ssd/MobileNetSSD_deploy.prototxt --model mobilenet_ssd/MobileNetSSD_deploy.caffemodel --input videos/<video-filename>.mp4 --output output/<video-filename>.avi --confidence 0.4 --skip-frames <frames-to-skip> --output-csv output/<video-filename>.csv --output-plots output/<video-filename>
  • example using video input:
python people_counter.py --prototxt mobilenet_ssd/MobileNetSSD_deploy.prototxt --model mobilenet_ssd/MobileNetSSD_deploy.caffemodel --input videos/ch05_20201211120000.mp4 --output output/ch05_20201211120000.avi --confidence 0.4 --skip-frames 60 --output-csv output/ch05_20201211120000.csv --output-plots output/ch05_20201211120000

Contributing

  1. Codes obtained from: https://www.pyimagesearch.com/2018/08/13/opencv-people-counter/

  2. Pre-trained model from: https://github.com/chuanqi305/MobileNet-SSD/tree/master/voc

  3. Primary results looks pretty promising. Logic goes like this: if u on the left of line and moving left, it classifies as going out and vice versa.

  4. Pros: Works fine, 0 training needed.

  5. Potential improvements:

  • find a pre-trained model with higher precision
  • fine-tune existing input parameters

License

Computer Center People Counter is released under the MIT License.