Uncertainty detection in images using Bayesian Convolutional Neural Networks

Purpose: To develop an image classifier using Bayesian Convolutional Neural Network to detect distracted drivers on a corpus of 22k+ drivers driving images with 10 different classes

Techniques used: CNN, Bayesian CNN

Tools Used: PyTorch, Jupyter Notebook, Python

• Baseline CNN - CNN_pytorch_baseline.ipynb & CNN_pytorch_baseline_0.ipynb

• Bayesian Neural Network - Bayesian NN.ipynb

• Bayesian Convolutional Neural Network - Bayesian_CNN.ipynb