Animal Classification Using Custom CNN Models

Animal Classification Using Custom CNN Models

Apr, 2024 - May, 2024 (Advisor – Prof. Prof. Nipun Batra, IIT Gandhinagar) | Project Link

• Developed and evaluated custom CNN models for animal classification across one-vs-rest, binary, and 5-class scenarios using a dataset of 90 animal images.
• Organized dataset for multi-class classification, implemented custom CNN architecture without pre-existing models, and utilized 3-fold cross-validation for robust evaluation.
• Achieved accurate classification results, visualized convolutional layer outputs to interpret learned features, and demonstrated effective model performance through classification matrices.