This repository contains the code and resources for a machine learning project aimed at detecting anomalies in industrial biscuit production. The project leverages the power of deep learning, utilizing advanced convolutional neural network (CNN) architectures like Xception and EfficientNet for image classification tasks.
https://www.kaggle.com/datasets/imonbilk/industry-biscuit-cookie-dataset
The project begins with a comprehensive exploratory data analysis (EDA) to understand the characteristics of the data. It then follows a structured machine learning pipeline including data preprocessing, model building, training, validation, and testing. Data augmentation techniques were applied to improve the model's robustness and ability to generalize.
Two CNN architectures were used: Xception, an advanced model known for its efficiency and performance, and EfficientNetB4, which is part of a family of models known for their excellent balance between accuracy and computational resource usage.
Comprehensive EDA and preprocessing of the Industrial Biscuits dataset. Use of Xception and EfficientNetB4 architectures for the classification task. Fine-tuning and comparison of the model performances. Use of data augmentation techniques for model generalization. Detailed record of the experimentation process and results. Strategies for further improvements and potential future work.
The EfficientNetB4 model emerged as the top-performing model, achieving perfect scores across all metrics. Despite the availability of larger and more complex models, the EfficientNetB4 model's efficiency and performance underscore its suitability for real-world, industrial applications.
This project demonstrates the potential of deep learning techniques in industrial quality control processes. We invite you to explore the repository, try the code, and improve upon our work. Contributions and suggestions are welcome!