The CIFAR-10 dataset is a collection of images of 10 different classes like cars, birds, dogs, horses, ships, trucks, etc. The idea of the project is to build an image classification model that will be able to identify what class the input image belongs to
Complete training and testing the model on cifar10 dataset can be seen in the attached .ipynb file, The architecture and design of convoulution neural network is made from scratch using pytorch backend. Testing accuracy of 78% is obtained on new data
In general accuracy of the model can be increased by tweaking some layers in convoulution network.