In the context of textile fabric, rare anomalies can occur, hence compromizing the overall quality. In order to avoid that, it is crucial to detect these rare defects.
Textile Defect Detection is a binary classification project which is able to predict whether the given patch of the fabric is damaged or not (good).
Dataset: Kaggle
- Importing the Dataset (HDF5 format).
- Pre-Exploration (Distribution of Classes).
- Preprocessig the dataset for binary classification.
- Exploring the dataset.
- Setting-up the model callbacks.
- Setting-up baseline metrics for the evaluation of actual models.
- Model building, training and evaluation (Transfer Learning).
- FineTuning the Model.
