CIFAR10

CIFAR10- Object Recognition

The repository contains implementation and evaluation of several Models for CIFAR10 Object recognition

Below are some Feature Extraction and Modules implemented.

1. Feature Extraction: 
        --> RGB 
        --> Standarized Image
        --> Edge Features.
        --> Histogram of oriented Gradients
        --> ZCA whitened

2. Models:
        --> K-nearest Neighbors
        --> Logistic Regression
        --> Support Vector Machines
        --> Deep Neural Networks
        --> Convolutional Neural Networks
        
3. Evaluation:
        --> Model Accuracy
        --> Confusion Matrix

Note: The majority of the code resides inside the MODEL folder. For Simplicity and deep understanding of several techniques/model, we emlploy and evaluate the models for only 2 classes (Airplane and Cat). However the code can be easily be extented for all the 10 labels, which would require a little bit of hyperparameter tuning.

Paper/Code References:

1. http://cs231n.github.io/convolutional-networks/
2. ImageNet Classification with Deep Convolutional Neural Networks - Alex Krizhevsky, Ilya Sutskever, Geoffrey E. Hinton
3. Maxout Networks : Ian J. Goodfellow David Warde-Farley Mehdi Mirza, Aaron Courville Yoshua Bengio
4. Dropout: A Simple Way to Prevent Neural Networks from Overfitting - Nitish Srivastava, Geoffrey Hinton, Alex Krizhevsky, Ilya Sutskever, Ruslan Salakhutdinov
5. Going deeper with convolutions - Christian Szegedy, Pierre Sermanet, Wei Liu , Yangqing Jia , Dumitru Erhan , Scott Reed , Dragomir Anguelov , Vincent Vanhoucke , Andrew Rabinovich