This research proposes a hybrid approach that integrates classical geometric techniques with neural network-based models to achieve robust and efficient camera localization. The goal is to develop a model that overcomes the limitations of both methods, enhancing robustness against noise, occlusions, and environmental variations. By combining the interpretability of classical methods with the adaptability of deep learning, this approach aims to improve localization accuracy and reliability, making it suitable for real-world applications in urban environments.

Самра А. (науч. рук. Шаветов С.В.) Hybrid (solved-learned) camera localization // Сборник тезисов докладов конгресса молодых ученых. Электронное издание. – СПб: Университет ИТМО, [2025]. URL: https://kmu.itmo.ru/digests/article/15033