This project classifies animal images using deep learning. The Animals with Attributes 2 dataset was preprocessed, and a CNN model was trained with TensorFlow. The model's performance was tested under different lighting conditions, and the impact of color stability techniques on accuracy was evaluated. Results were analyzed on Kaggle.