Pruthvi-Sanghavi / transfer_learning

Digit Recognition and Transfer Learning

Digit Recognition and Monkey classification.

Libraries

• Tensorflow
• Matplotlib
• Keras
• Numpy

Editor used

• Jupyter Notebook

Setup Instructions (for windows)

• Unzip the compressed folder
• Open command prompt
• Navigate the code directory

cd pruthvi_proj2/pruthvi_P2/code/

• Run the following.

pip install jupyter notebook
pip install -r requirements.txt

Datasets

• MNIST Dataset
• Monkey Dataset link

For Monkey Dataset

• Download the dataset, extract the files

cd code/monkey_classifier
mkdir datasets

• Copy the training, validation folders and monkey_labels.txt from the downloaded folder and paste it to datasets folder.

Training the model

• Type the following in command prompt
• For Digit Recognition using CNN

cd code/digit_cnn
ipython notebook digitCNN.ipynb

• For Digit Recognition using SVM

cd code/digitSVM
ipython notebook digitLSVM.ipynb
OR
ipython notebook digitKPSVMSVM.ipynb
OR
ipython notebook digitKRSVM.ipynb

• For Monkey Classification using CNN

cd code/monkey_classifier
ipython notebook monkeyCNN.ipynb

• For Monkey Classification using Transfer Learning

cd code/monkey_classifier
ipython notebook monkeyTL.ipynb

Using the Trained models

cd code/monkey_classifier
ipython notebook monkeyTL_trained.ipynb

cd code/digit_cnn
ipython notebook digit_CNN_trained.ipynb