Framework documentation
Foundations
The framework will take care about processing the image frame(s) and detect the existence of violence. Simple output yes or no.
It is composed by python detection class file and a frozen model file.
The target is being as much as possible platform invariant.
Types of violence targeted
-Assault
-Fight
-Lynching
-Weapon identification ( gun, knife...)
-Terrorism
-Violent behavior
Scope
It's a real time violence detector framework based on computer vision.
It comprehends but not limited to:
-Dataset and it's composition
-Image preprocessing
-Network architecture: object detection / image classification
-Keypoint extraction
-Decision rules for the detection
*Software must be hardware independent as much as much as possible
It is not:
-A final product development
-Hardware related project
-Privacy related discussion
-It did not supposed to cover:
-Software not related to the detection
-OS related matters as drivers, packages or external dependencies
We believe that technology can make the world safer.
Full details available in Github project repo.
contact@violencedetector.org