This little project should detect all persons in an image/video and mark everybody red (unmasked) and blue (masked). It can help stores to check if their clients wear masks during the corona-pandemic!
 |
 |
|---|---|
| Unmasked Person | Masked Person |
You have to give at least one argument to run this code. There are some available:
-helpThat is clear-testTest images inside 'data/test' and returns the result per image-webcamGet a videostream of a webcam and check for faces-train [epochs]Train images. Epochs has to be given-videoTest and/or train entire videos
First you have to define a webcam in the webcam.json file. If you don't own a webcam, you can simply download an IP Webcam app. I am using this one: IP Webcam (Android)
After downloading, you can start it and press the button 'Start server' at the bottom. On your screen should something be visible like http://192.168.178.38:8080. Enter this link in a browser and you see the livestream of your mobile camera. Then open the webcam.json file. Paste the link and add /video. The reason is that we only want the videostream. At the end you can name it as you want.
[ { "name" : "mobile_camera", "url" : "paste_here_your_link + /video" } ]
You can add more webcams easily: Copy and paste this ,{ "name" : "mobile_camera", "url" : "paste_here_your_link + /video" }. NOTE: You have to give each cam another name. Otherwise there will be only on window.
If you done everything correctly you can double click Launch Webcam or run python main.py -webcam in a cmd.