/iVision-Data-Collector-API-1

This is an attempt to collect as much as possible photos of a disguised human which will be used for training an AI model.

Primary LanguagePython

iVision-Data-Collector-API-1

This API Enables you to collect images when a motion is triggered. The input feed is given from IP Cam or Device Cam.

The gathered images could be used for training your ML Model.

An Android APP called DroidCAM runs in your mobile phone making your phone to act as IPCAMERA, your mobile phone is connected to your home Wi-Fi router.

A custom developed windows application called IVISION DATA COLLECTOR runs in your laptop which is connected to the home Wi-Fi network. The mobile phone is also connected to the laptop which is in same home Wi-Fi network. This IVISION DATA COLLECTOR application has inbuilt motion detection capability with which starts to store the JPEG files in a defined folder. These JPEG files are the DATA which will be further pre-processed, labeled before it is used as training dataset for the AI model. All we need is to do this setup and start acting (like walking in disguise) in the front of the mobile phone camera and use the windows application to collect that data. And upload those collected jpeg files into the common Google drive. Use your imagination to disguise and start taking the data!

Some tips:

  1. Use a stable place (like a tripod.) to hold the mobile phone without shaking.
  2. Look for any constant moving items (like celling fans, fluttering paper etc..) in the video frame and avoid them as it will un-necessarily trigger the motion detection
  3. You can try to take diversified data different angles, different lighting conditions, with children in disguise, with multiple human bodies disguised together (say two humans sharing a bedsheet to disguise and walking across) Running across the camera frame in disguise Crawling across the camera frame in disguise
  4. It will be better to provide at least 400 to 500 Snapshots per scenario for proper AI training
  5. Try to give a much as possible diversified scenarios for a better AI training

Steps to execute:

  1. Install DROIDCAM APP (free version from Playstore) in your Android mobile phone (this will convert your mobile phone to a wireless web camera)
  2. Connect your Android phone to your home Wi-Fi router.
  3. Download iVision Data Collector 1.1.1.exe to your Windows Laptop
  4. Install the IVISION DATA COLLECTOR windows application in your laptop.
  5. Connect your Windows LAPTOP to your home Wi-Fi router.
  6. Click iVision Data Collector 1.1.1.exe in the directory in which you have installed the application to invoke the application
  7. Click DROIDCAM app in your Android Phone
  8. Use IVISION DATA COLLECTOR inbuilt TEST CAMERA option to test camera and laptop video streaming connectivity.
  9. Create Multiple (say one folder for one scenario) directories in your Laptop 10.Use the IVISION DATA COLLECTOR tool to choose the directories one by one and start taking the data, while acting in the front of the mobile phone camera with various disguise