Readme.md

Image classification for classifying cats and dogs

Image classification of for Kaggle cats and dogs challenge based on tutorial on Keras web blog.

A ConvNet model following the tutorial at the blog. The model parameters were further saved to load the model and perform sentiment analysis.

Data source: Kaggle Cats vs Dogs.

Future prospects:

1. Improve model with different algorithms.
2. Try more pre-trained image classification models.

Folder structure:

1. Root: Main folder containing Readme and scripts.
2. data: Data folder containing:
   a) train: Training data.
   b) validation: Test/Validation data.
   c) test: Test data.
3. figures: All generated figures from the scripts.
4. models: Trained models stored in this folder.

Files

utilities.py: Contains useful functions.
ImageClassification.py: ImageClassification file.
CheckImages.py: Image augmentation example.

File descriptions

=======================================================
utilities.py: This script is required for all other scripts.

Useful common functions.

=======================================================
ImageClassification.py:

Image classification using ConvNet neural networks.

With simple network validation accuracy are 75%.
Large fluctuations in accuracy history.

Loss function and Accuracies are plotted in Figures:
Accuracy.png
Loss.png