Burhanuddin Rangwala - 1911109
Mihir Pandya - 1911098
In this project we create a robust image classification pipeline to classify images.
We created 2 different models and recorded their accuracy and loss.
The app has been deployed at https://share.streamlit.io/bamblebam/cifar-10-image-classification/app.py
The dataset used in this project is the CIFAR-10 dataset which consists of 60000 32x32 colour images in 10 classes, with 6000 images per class. There are 50000 training images and 10000 test images.
This is a basic CNN model which uses 2D convolution layers with an increasing amount of filters. Max pooling and Batch normalization are used as regularization techniques. The total number of parameters is ~4.4 million.
This is the novel ConvMixer model which uses image patches which are synonymous to word embeddings in NLP as input. The model uses a conv-mixer layer which is nothing but a residual depthwise convolution followed by a pointwise convolution i.e. depthwise separable convolution. This layer is repeated a certain number of times. The total number of parameters in this model is ~1.1 million.
