Bridging AI and wildlife conservation: Classifying wildcats from genetically validated images

This study presents a multidisciplinary approach for classifying three cat categories, European wildcats ( Felis silvestris , WC), hybrids ( F. silvestris × catus , HY), and domestic cats ( F. catus , DC), using convolutional neural networks (CNNs) on real-world images collected from camera traps, roadkills, and research teams. The initial dataset included 3,392 images (WC = 1,738; HY = 461; DC = 1,193) and was filtered using YOLOv12 to retain clearly visible cats, resulting in 2,518 images (WC = 1,201; HY = 461; DC = 856). We achieved the highest CNN’s accuracy (86%) on the balanced dataset and tested on a genetically validated dataset, 71% on the genetically validated dataset. When trained on the larger, unbalanced dataset, accuracy on the genetic test increased to 81%, indicating that enlarging datasets may effectively compensate for class imbalance. An offline-capable mobile app was developed for field use, supporting local photo capturing and usability feedback via an included and structured questionnaire for future analysis. This work is the result of a synergistic collaboration among computer scientists, wildcat researchers, and citizen scientists. The proposed model offers a consistent and reproducible field-ready tool for European wildcat conservation. • A novel dataset of wild-living cats, including genetically validated images. • A lightweight CNN–Transformer model achieves up to 81% accuracy on genetic test data. • A mobile app for offline classification of wildcats, hybrids, and domestic cats in field conditions. • Validation against genetic analyses ensures reliable and reproducible classification outcomes. • Public release of pre-trained models, code, and dataset to foster reproducibility. • Integration of AI methods, software implementation, and citizen science for conservation support.