People Count

A module for counting people in a picture based on Yolo v.3 & S-DCNet

Pre-requisite

• Install all the required packages listed in the requirements.txt

  $ pip install -r requirements.txt
  

• Download pre-trained weights file weights.tar.gz

  • Google Drive
  • Extract the downloaded weight file inside the project folder

Test

• Run detect.py module with image_folder information

  $ python detect.py --image_folder /path/to/images
  

• To save result in a log file

  $ python detect.py --image_folder /path/to/images \
                     --log_enable 
  

  -- output.log file will be created and all the prediction result will be saved there

• To save result in a db

  $ python detect.py --image_folder /path/to/images \
                     --db_enable \
                     --db_host localhost \
                     --db_name DB_NAME \
                     --db_user_name DB_USER_NAME \
                     --db_user_pw DB_USER_PW \
                     --db_table DB_TABLE_NAME
  

  • Assuming the schema of the DB table as follows

  # Example Schema
  CREATE TABLE `count` (
               `filename` varchar(45) NOT NULL,
               `htag` varchar(45) DEFAULT NULL,
               `model` int(11) DEFAULT NULL,
               `result` varchar(255) DEFAULT NULL
               ) ENGINE=InnoDB DEFAULT CHARSET=latin1;