This paper explores the use of deep learning algorithms for anomaly detection on small wind turbine blades, aiming to improve maintenance efficiency and reduce costs. The authors investigate five base deep learning architectures—Xception, ResNet-50, AlexNet, VGG-19—and a custom convolutional neural network (CNN). Transfer learning approaches are also proposed and developed, utilizing these architectures as feature extraction layers. A new dataset containing 6000 RGB images of healthy and damaged small wind turbine blades is created, including both indoor and outdoor images. Each model is tuned using different layers, image augmentations, and hyperparameter tuning to achieve optimal performance. The results show that the proposed Transfer Xception outperforms other architectures, achieving 99.92% accuracy on the test data. Additionally, the performance of the models is compared on a dataset containing faulty and healthy images of large-scale wind turbine blades, where the Transfer Xception achieves 100% accuracy. These findings highlight the potential of machine learning in wind turbine blade fault identification.This paper explores the use of deep learning algorithms for anomaly detection on small wind turbine blades, aiming to improve maintenance efficiency and reduce costs. The authors investigate five base deep learning architectures—Xception, ResNet-50, AlexNet, VGG-19—and a custom convolutional neural network (CNN). Transfer learning approaches are also proposed and developed, utilizing these architectures as feature extraction layers. A new dataset containing 6000 RGB images of healthy and damaged small wind turbine blades is created, including both indoor and outdoor images. Each model is tuned using different layers, image augmentations, and hyperparameter tuning to achieve optimal performance. The results show that the proposed Transfer Xception outperforms other architectures, achieving 99.92% accuracy on the test data. Additionally, the performance of the models is compared on a dataset containing faulty and healthy images of large-scale wind turbine blades, where the Transfer Xception achieves 100% accuracy. These findings highlight the potential of machine learning in wind turbine blade fault identification.