CNN-Based Classification of Feline Facial Behavioral Signals Using Augmented Datasets

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

In this work, an automatic recognition system for cats' facial expressions is developed using a Convolutional Neural Network (CNN) that can classify cat facial expressions into seven observable behavioral display categories: Agonistic Open-Mouth Display, Retching/Gagging Display, Relaxed Open-Mouth Display, Neutral Expression, Low-Arousal Passive Display, Fear/Flight-Readiness Display, and Orienting/Alerting Display. An initial set of 511 unique cat pictures was used to create a training dataset of 4,344 instances after applying ten types of data augmentation, such as horizontal flipping, vertical flipping, rotation, blurring, brightness adjustment, noise addition, grayscale conversion, and cropping. The proposed CNN model consists of two convolutional layers using 5 × 5 filters, two max-pooling layers, and two fully connected layers with dropout. The obtained classification accuracy ranges from 90.50% (Neutral Expression) to 98.61% (Agonistic Open-Mouth Display). The system achieves an overall weighted accuracy of approximately 95.30%.