Project information
- Category: Deep Learning Classification
- Course: Machine Learning Zoomcamp
- Project date: 26 December, 2021
- Project Repo: https://github.com/AbdassalamAhmad/Predicting-Clothing-Types
- Project Live Demo: https://share.streamlit.io/abdassalamahmad/predicting-clothing-types/main/streamlit_DLapp.py
Project Details
Predicting 10 different clothing types [dress, hat, longsleeve, outwear, pants, shirt, shoes, shorts, skirt, t-shirt]
using Xception pre-trained model from Keras library by using transfer learning.
Here are the main steps I did to complete this project:
- Freeze Xception CNN layers and train its dense layers.
- Hyperparameter tuning: Changed the parameters of the model like (learning rate, number of epochs, size of dense layers..etc.) to achieve better accuracy.
- Data Augmentation : created multiple images from a single image to generalize the model.
- Evalutaing the model, achieved 90% accuracy.
- Deployed the model to Streamlit for everyone to test it.