Personal Protection Equipment Detection based on Deep Learning

Real time Personal Protection Equipment(PPE) detection running on NVIDIA Jetson TX2 and Ubuntu 16.04

• Person, HardHat and Vest detection
• Input from Video file or USB Camera
• A backend service which can push message to "console" or "Cisco® Webex Teams space" when an abnormal event is detected.

Requirements

• NVIDIA Jetson TX2 or Ubuntu 16.04
• NVIDIA GPU on Ubuntu 16.04 is optional
• Python3

How to run

Video Inference Service

$ cd inference
$ pip3 install -r requirements.txt
$ python3 video_demo.py --model_dir=xxx  --video_file_name=xxx --show_video_window=xxx --camera_id=xxx

• model_dir: the path to model directory
• video_file_name: input video file name or usb camera device name, you can get camera device name on ubuntu or NVIDIA Jeston by running

$ ls /dev/video* 

• show_video_window: the flag to show video window, the options are {0, 1}
• camera_id: It is just convenient for humans to distinguish between different cameras, and you can assign any value, such as camera001

Backend Service

run the following command

$ cd backend
$ pip3 install -r requirements.txt
$ python3 main.py

run application as docker

docker-compose up
or
docker-compose up --build

send notification

By default, it will use the console notification, this just print the notification to stdout. If you want to use Cisco® Webex Teams, use change the config referring to config.py. Or you can write your own if you write your provider inheriting the notification.Provider

setup Cisco® Webex Teams

• create a robot referring to https://developer.cisco.com/webex-teams/, you will get the token
• create a webex-teams room and add the robot to that team
• go to https://developer.webex.com/docs/api/v1/rooms/list-rooms to get the new created room id
• put the above info to the config.py

Alert Message Format

• total_person: number of people detected
• without_hardhat: number of people without hard hat
• without_vest: number of people without Vest
• without_both: number of people without hard hat and vest

Training Program

Based on TensorFlow Object Detection API, using pretrained ssd_mobilenet_v1 on COCO dataset to initialize weights.

Training Data

coming soon!

Reference work

• TensorFlow Object Detection: https://github.com/tensorflow/models/tree/master/research/object_detection