Plant Leaf Classification Dataset: A Labeled Collection for Deep Learning-Based Image Recognition
Description
This dataset contains 12 class-high-resolution images of plant leaves, mainly collected for machine/deep learning-based classification. The 12 classes are: Artocarpus heterophyllus,Bambusa vulgaris,Coccinia grandis,Codiaeum variegatum,Ficus benghalensis,Foliorum Forma-Orbicularis,Herbarium Oxalis,Hevea brasiliensis,Indonesian Bay,Lembu,Mangifera indica,Psidium guajava. Each class contains approximately 300 images and all over almost 3700 images in total 12 classes. This image is resized (512x512); before using this dataset, you need to apply preprocessing and augmentation. This dataset
Files
Steps to reproduce
Brief Note on Reproducing the Process Download the Dataset – Ensure the dataset is organized into train, val, and test folders, with images sorted into class-wise subfolders. Set Up the Environment – Install required dependencies (TensorFlow, NumPy, Matplotlib, Seaborn, Scikit-learn). Load the Dataset – Use ImageDataGenerator to preprocess images and create data loaders for training, validation, and testing. Define and Train CNN Model – Implement a sequential CNN model with convolutional, pooling, and dense layers, and train it for 25 epochs. Evaluate the Model – Test the model on unseen data and compute accuracy. Visualize Performance – Plot accuracy trends across epochs and generate a confusion matrix for performance analysis. Save the Model – Save the trained model (plant_leaf_cnn.h5) for future use. These steps ensure easy reproduction of dataset usage and model training.
Institutions
- Daffodil International University