Lithofacies identification plays a pivotal role in the realm of oil and gas exploration and reservoir characterization. Presently, machine learning-based lithofacies identification leveraging well log curves has emerged as the predominant approach. However, the efficacy of machine learning methodologies is often contingent upon access to a large amount of well log data and corresponding annotations for training. Yet, in practical scenarios, the availability of drilling and core samples is typically limited, thereby posing significant challenges to the precision and generalization capability of various machine learning models. To address these challenges, we propose a novel transfer learning methodology that integrates elements of both classical neural networks and quantum neural networks. This innovative approach entails initial pre-training of the model utilizing classical neural network architectures, followed by the integration of quantum neural networks for subsequent fine-tuning. This amalgamation serves to augment the generalization capability of the pre-trained model. Subsequently, we validate the efficacy of our proposed quantum transfer learning methodology utilizing authentic well log data. Comparative analyses with conventional transfer learning methods reveal that our model exhibits superior lithology identification accuracy and heightened generalization prowess. Our proposed approach holds promise for offering valuable insights into lithofacies identification tasks within regions characterized by limited core sample availability.