Vaginal cytology is a diagnostic tool for evaluating estrous cycle stages and reproductive health in female dogs and cats. It involves microscopic examination of vaginal epithelial cells, but subjective interpretation can lead to inconsistencies. This study explores artificial intelligence (AI), specifically deep learning, to enhance accuracy. A total of 1,096 vaginal smear samples were collected, stained, digitized, and analyzed using AI. Several pre-trained convolutional neural networks (CNNs), including MobileNetV2, ResNet152V2, EfficientNetV2L, Xception, VGG-16, InceptionV3, NasNetLarge, InceptionResNetV2, DenseNet201, and ConvNeXtSmall, were evaluated. The Xception model achieved the highest accuracy at 97.65%. These findings demonstrate AI’s potential to reduce subjectivity, improve diagnostic consistency, and advance reproductive health assessments in veterinary medicine.