Description

Computer Vision classification for violence using a VGG16 Architecture with pictures and videos.

Details

• VGG16 Architecture: Three Convolution2D layers with a FC dense layer as the output.
• Libraries: TensorFlow Keras and Sklearn for the model. CV2 and OS for data preprocessing.
• Image Prediction: Predicts violence for multiple images. Graphs their predictions with the original images using MatPlotLib.
• Video Prediction: Predicts violence for each frame of a video.
• Video Blurring: Capable of blurring violence detected frames of videos then writing the altered version to a new .mp4v
• Individual Project
• One Month Scope