Animal Species Classification Dataset

Dataset Composition:

The dataset is structured into six key directories, each serving a specific purpose within the machine learning pipeline:

1. Interesting Data:
   • This directory contains 5 unique and challenging images per species class. These “interesting” images are selected to test the model’s ability to make accurate predictions in complex scenarios. Evaluating model performance on these images offers insights into its understanding and classification capabilities.
2. Testing Data:
   • A randomly populate directory with images from each species class, specifically curate for model testing. This dataset is essential for evaluating the performance and generalization of the model after it has been train.
3. TFRecords Data:
   • This directory includes the dataset formatted as TensorFlow records. All images have been preprocessed, resized to 256×256 pixels, and normalized. These ready-to-use files facilitate seamless integration into TensorFlow-based machine learning workflows.
4. Train Augmented:
   • To enhance model training, this directory contains augmented versions of the original training images. For each original image, 5 augmented variations are generated, resulting in a total of 10,000 images per species class. This augmentation strategy is crucial for increasing dataset size and diversity, which in turn helps the model learn more robust features.
5. Training Images:
   • This directory is dedicated to the core training data, with each species class containing 2,000 images. The images have been uniformly resized to 256×256 pixels and normalized to a pixel range of 0 to 1. These images form the backbone of the dataset, enabling the model to learn distinguishing features for each species.
6. Validation Images:
   • The validation directory contains 100 to 200 images per species class. These images are used during the training process to monitor the model’s performance and adjust hyperparameters accordingly. By providing a separate validation set, the dataset ensures that the model’s accuracy and reliability are rigorously evaluate.

Species Classes:

The dataset includes images from the following 15 animal species:

1. Beetle
2. Butterfly
3. Cat
4. Cow
5. Dog
6. Elephant
7. Gorilla
8. Hippo
9. Lizard
10. Monkey
11. Mouse
12. Panda
13. Spider
14. Tiger
15. Zebra

Each class is carefully curated to provide a balance and comprehensive representation of the species, making this dataset suitable for various image classification and recognition tasks.

Application:

This dataset is ideal for building machine learning models aim at classifying animal species. It serves as a valuable resource for academic research, conservation efforts, and applications in wildlife monitoring. By leveraging the diverse and augment images, models train on this dataset can achieve high accuracy and robustness in real-world classification tasks.