An effective way to implement social distancing for On-Site Service employees working at places like warehouses, malls etc
where the employees cannot be shifted to work from home.

• Python
• OpenCV
• Keras
• Tensorflow
• Yolov3
• Flask

All the dependencies and required libraries are included in the file requirements.txt See here

1. Clone the repo

$ git clone https://github.com/kaurramanpreet/Myntra_Comp.git

2. Change your directory to the cloned repo and create a Python virtual environment named 'test'

$ mkvirtualenv test

3. Now, run the following command in your Terminal/Command Prompt to install the libraries required

$ pip3 install -r requirements.txt   (if python 3 is installed)
$ pip install -r requirements.txt   (if python 2 is installed)

1. Open terminal. Go into the cloned project directory and type the following command:

$ python app.py

2. Open browser. Type localhost:5000 in address bar and press enter

• app.py is the root of the web application. This python script will link all the features of web-app.
• Code for hosting the full project as a web application is implemented in WebServer folder.
• Code for counting the number of people entered and exited the area has been added in ENTRY_EXIT COUNT folder.
• Code for detecting nose and mask i.e. mask has been worn, not worn and not properly worn has been added in MASK AND NOSE DETECTION.
• Code for measuring the distance between 2 people is implemented which keeps a check on the social distancing rules and had been added in distance folder.
• requirements.txt contains every installed dependency of our porject.

• This is the root of our web application.This is where all the Flask application goodness will go.It contains routes of all the features along with the functionality to run them.

• currently the work includes to run all the features of our project from a web application.
• static folder contains all the output gifs and images .
• templates folder contains all the html pages that are being loaded on the web application.

• counter.py :- The file that contains the implementation logic of people counter
• people_capture.mp4 :- The input video file used to test the code

for checking this code just run the counter.py file and you would be able to see the count of number of people entered in a store(room) and the count of people exited from the store(room) 

• Initially the work includes to check whether the person is wearing the mask or not

• So we have tried to include the other part as well in which even if the person is wearing the mask , our code will check whether the mask is worn properly or not by detecting the nose(nostrils) or if the person is using hands instead of mask.

• code_mask.py :-The file that contains the implementation logic of Mask Detection.
• haarcascade_frontalface_alt2.xml :- Pre-trained haarcascade model to detect face.
• haarcascade_mcs_nose.xml :- Pre-trained haarcascade model to detect nose.
• mask_v1.mp4 :- The input file used to check the code for the validity in the above 3 cases.
• mask_v2.mp4 :- The input file used to check the code by using hands instead of mask.
• model.h5 :- Pre-trained MobileNet-V2 model based on ImageNet Database
• Dataset Link:- DATASET

for checking the code just go to this folder and run the code_mask.py file and the results would be shown as

• it is used to find the distance between 2 persons and generated an red alarm when the people are not following the social distancing rule
• SocialDistancingDetector.py :- The file that contains the implementation logic of checking the social distance between people.
• vtest.avi :- The input file used to check the code.
• yolo
  • coco.names :- 80 names of objects (labels) that can be detected on the image.
  • yolov3.cfg :- Configuration file with complete information about number of nodes in layers .
  • yolov3.weights:- values of neurons i.e weights already trained are stored in this file.
    • Link to weights file :- WEIGHTS

for checking the code just go to this folder and run the SocialDistancingDetector.py file and the results would be shown as