8.1 Fashion classification

8.1 Fashion classification

Slides

If you have problems cloning the repository with the data, use a different command for cloning:

git clone https://github.com/alexeygrigorev/clothing-dataset-small.git

In this session, we'll be working on multiclass image classification with deep learning. Some deep learning frameworks like TensorFlow and Keras will be implemented on clothing dataset to classify images of clothes as t-shirts, hats, pants, etc.

The dataset has 5000 images of 20 different classes. However, we'll be using the subset which contains 10 of the most popular classes. The dataset can be downloaded from the above link.

Useful links:

• Full dataset: https://www.kaggle.com/agrigorev/clothing-dataset-full
• Subset: https://github.com/alexeygrigorev/clothing-dataset-small
• Corresponding Medium article: https://medium.com/data-science-insider/clothing-dataset-5b72cd7c3f1f
• CS231n CNN for Visual Recognition: https://cs231n.github.io/

Notes

Add notes from the video (PRs are welcome)

• introduction to this weeks topics
• using images as data source instead of tabular data
• classifying images of t-shirts using a convolutional neural network
• 5000 images and 20 different classes https://www.kaggle.com/agrigorev/clothing-dataset-full
• see corresponding medium.com article https://medium.com/data-science-insider/clothing-dataset-5b72cd7c3f1f
• deep learning with tensorflow and keras
• parameters, layers, regularization

• Notes from Peter Ernicke

Navigation

• Machine Learning Zoomcamp course
• Session 8: Neural Networks and Deep Learning
• Next: Setting up the Environment on Saturn Cloud