The DataSet is divided into training, testing and val sets,will combine them and randomly re-divide
Main task is to understand and implement the basic Image Classification pipeline that can correctly predict the label of a given image and conduct the following experiments:
-
Understand the train/test splits. Even if the dataset is divided into training and test sets, combine them both and randomly re-divide them with at least 75% of data used for training.
-
Implement sophisticated methods of feature generation (a minimum of 8 features are required for each instance)
-
Predict the outcome of the model for each of the following experiments
- Train my model using any classification algorithm
- Fewer features for each image, compare my model when using all features, half of them and then only 2 features.