Binary Cat–Dog Classifier is a deep learning project that demonstrates an efficient approach to binary image classification using modern convolutional neural networks. The model is trained on the Oxford-IIIT Pet Dataset, leveraging transfer learning with a ResNet-34 backbone to accurately distinguish between cat and dog images.

This project includes the full pipeline: dataset preparation, exploratory visualization, model training, performance evaluation, and prediction on external images. With only four fine-tuning epochs, the classifier achieves over 99% accuracy, making it a strong example of how transfer learning can produce high-performance models even in small-scale experiments.

Additionally, the notebook provides reproducible setup steps, deterministic validation splits, and clean model export, making it suitable as a template for future image recognition tasks.