Skip to content

Detect motion for input video source using background subtraction algorithm. Then dynamically extract the images from the video source as test images. This images are fed to trained model "64x3-CNN.model" to predict if the motion is caused by Humans or not. Then display alert message on Security feed.

Notifications You must be signed in to change notification settings

santos97/CNN-Keras-TF-Surveillance-System

Repository files navigation

CNN-Keras-TF-Surveillance-System

Detect motion for input video source using background subtraction algorithm. Then dynamically extract the images from the video source as test images. This images are fed to trained model "64x3-CNN.model" to predict if the motion is caused by Humans or not. Then display alert message on Security feed.

Output

Night Vision vs Thermal:

Object under observation is behind bushes

Objects viewed from sky

Background Subtraction:

Screenshot 2019-05-21 at 10 03 21 PM

Frame Dialation:

Screenshot 2019-05-21 at 10 03 38 PM

Final Security Feed:

Screenshot 2019-05-21 at 10 04 03 PM

About

Detect motion for input video source using background subtraction algorithm. Then dynamically extract the images from the video source as test images. This images are fed to trained model "64x3-CNN.model" to predict if the motion is caused by Humans or not. Then display alert message on Security feed.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages