CS175-Distracted-Driver-Detection #RESNET18 #VGG16 #MOBILENET #PYTORCH #MACHINELEARNING
Classification of various distracted driver poses using CNNs such as VGG16 and Mobilenet models.
Distracted Driving:
The impact from a vehicular accident is enormous, yet totally preventable. To avoid these types of tragedies we wish to develop an autonomous detector that can predict whether or not a driver is distracted through the use of dash cams.
Datasets: https://www.kaggle.com/c/state-farm-distracted-driver-detection/data
Training data: Contains 22,424 images in the training data separated into 10 labeled class folders.
Testing data: 79.7 thousand unlabeled images (tested for accuracy through kaggle submission).
Size of each image given to us is 640 × 480 pixels.
Here's a comprehensive explanation of what we did and our report too: https://docs.google.com/document/d/16JsjBL2qVtuH3BgetJh-_jYUY93D5nVEIc6CwqqScik/edit?ts=60bea8fb
To follow the project start with the notebook ..., and follow along the code