Flood detection using CNN
Basic Implentation of CNNs to detect Floods from images.
First is the CNN implementated from scratch and then second is using the VGG-16 pretrained network. CNN without using pretrained network gives 50% accuracy and with VGG-16 gave 70% accuracy.
There are 3 folders: train, test and valid. Each consist of separate sub-folders Floods and Non Floods.