- Choose an image dataset from your choice ✅
- Prepare data preprocessing & cleaning techniques of your choice according to the nature of your dataset ✅
- Split data to train, validation and test data ✅
- Implement 4 different architectures (try different layer depth, drop out, hyperparameters)
- Choose the correct loss function and optimizer ✅
- Display model summary for each architecture ✅
- Create predict() function that takes a test sample and prints the class of that test sample. ✅
- Compare the result of all 4 architectures and choose the best one