Abstract
This study evaluates the effectiveness of the YOLOv11 Nano neural
network architecture for dog breed classification. YOLOv11 Nano is
a small convolutional neural network (CNN) model specifically designed
for image classification task, particularly suited for objects
with similar characteristics, such as dog breeds. The proposed architecture
presents results comparable to previously applied networks,
such as NASNet-A Mobile, and slightly underperforms in comparison
to Inception-ResNet-V2, despite its substantially smaller model size. A
comprehensive evaluation was conducted, incorporating pre-trained
models alongside optimization strategies, including data augmentation
and K-Fold cross-validation. The results highlight the potential of
YOLOv11 Nano for the dog breed classification task, offering a balance
between accuracy and size.