Flower Classification Project

Author: Sevendi Eldrige Rifki Poluan

Project Overview

This project trains a Convolutional Neural Network (CNN) to classify flower images into 5 categories:

• Sunflowers (699 images)
• Tulips (799 images)
• Dandelions (898 images)
• Roses (641 images)
• Daisies (633 images)

The dataset used is the TensorFlow flower_photos dataset. The main objective is educational: to practice the full deep learning workflow, from data loading and preprocessing to model training, evaluation, and confusion matrix analysis.

Workflow Summary

The notebook flower.ipynb follows this pipeline:

1. Import required libraries.
2. Download and inspect the dataset.
3. Build train/validation datasets using TensorFlow dataset utilities.
4. Define and compile a CNN model.
5. Train the model with early stopping.
6. Save and reload the model.
7. Evaluate performance and plot metrics.
8. Generate a confusion matrix.

Tools and Libraries

• TensorFlow / Keras
• NumPy
• Pandas
• Matplotlib
• scikit-learn

Figures

Sample Flower Images

This figure shows sample inputs from the dataset.

Training History

This figure shows the model training history (for example, loss/accuracy trends).

Confusion Matrix

This figure shows class-wise prediction performance.

References

Dataset and Documentation

1. TensorFlow Flower Photos Dataset (download link used in this project): https://storage.googleapis.com/download.tensorflow.org/example_images/flower_photos.tgz
2. TensorFlow API: image_dataset_from_directory: https://www.tensorflow.org/api_docs/python/tf/keras/preprocessing/image_dataset_from_directory
3. TensorFlow Tutorial: Image Classification: https://www.tensorflow.org/tutorials/images/classification

Related Papers

1. Krizhevsky, A., Sutskever, I., and Hinton, G. E. (2012). ImageNet Classification with Deep Convolutional Neural Networks. NeurIPS. https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf
2. Simonyan, K., and Zisserman, A. (2015). Very Deep Convolutional Networks for Large-Scale Image Recognition (VGG). ICLR. https://arxiv.org/abs/1409.1556
3. He, K., Zhang, X., Ren, S., and Sun, J. (2016). Deep Residual Learning for Image Recognition (ResNet). CVPR. https://arxiv.org/abs/1512.03385

Notes

• This repository focuses on learning and experimentation.
• For stronger performance, future versions can compare this custom CNN with transfer learning models (for example, MobileNetV2 or ResNet50).