This is an image classification project based on the past Kaggle Competition.
- Keras to build Neural Network Model.
- Other required imports are very general to any Machine Learning project (like Numpy and Matplotlib). You can surely follow along with the code file.
This an image classification project with almost 25000 images used for training. Consider using a GPU for training purpose. You can use Kaggle Kernels or Google Colaboratory as well.
The training has been done twice, once with just basic preprocessing (converting images to pixel tensors) and again with image augmentation. Try to reproduce the results by running in Jupyter Notebook. Upload your results in Kaggle and try to improve.
- Try using a pre-trained network to extract the features and then your own dense layer for classification. Compare the results.