Implementing CNN Architectures

In this repository you will find the implementation of ResNet and Inception-ResNet from scratch and also a CNN model for classifying natural disaster images.

ResNet Implementation with Keras

In this notebook, resnet is implemented using keras and trained on cifar100 dataset.

Inception-ResNet Implementation with Keras

In this jupyter notebook, inception-resnet is implemented using keras and is trained on cifar100 dataset.

Classification of Images using Keras

Natural Disaster Classification with Keras

Classification of natural disaster images into four classes which are:

1. Cyclone
2. Wildfire
3. Flood
4. Earthquake

The classification model is created as a convolutional neural network (CNN) using keras layers. You can find how to:

• load data and split into train and test sets
• load data in batches using data generators when the dataset is large
• create a CNN model from scratch and train on your dataset
• evaluate the model's performance and view confusion matrix
• plot the learning curve