Week 3 journal

Week 3 milestone:

Consist of already having a complete training set data and completed feature, and working on building the model database with features.

Comparing your progress to the milestone:

Since I have been working on my training dataset during Winter break, the dataset was done by end of Week 1 with 52 different classes which are the labels of images that belong to that class. Also, the dataset is already loaded and preprocessed to Colab and ready for training my CNN model. In week 3, I have been working on the architecture of my CNN, and the best architecture of my CNN that I have got is 84.2% of evaluate accuracy to test data. My week 3 milestone

What I have learned so far:

In the past 3 weeks, I have learned how to do data preprocessing, converting raw data into a format suitable for training a CNN, including resizing, normalizing and converting images to tensors. I have also learned data splitting, dividing the data into training, validation, and test sets. In building CNN model, I have learned dropout to prevent overfitting.

Week 7 Milestone:

By week 7, the convolutional layers are complete and my algorithm will be able to detect playing card on colab and H7 broad. In the next few weeks, I will mainly work on the H7 broad part.