DeepBrownConrady: Prediction of Camera Calibration and Distortion Parameters Using Deep Learning and Synthetic Data

This paper introduces DeepBrownConrady (DBC), a novel deep learning model that predicts horizontal field of view (H-FOV), principal points, and Brown-Conrady distortion parameters from single input images.

Key contributions include:

• A synthetic data pipeline that generates large-scale training datasets for calibration.
• Robust prediction of intrinsic camera parameters without requiring calibration targets.
• Validation on both synthetic and real datasets, demonstrating strong generalization.

The paper highlights the potential of synthetic data in enabling scalable camera calibration for robotics, AR/VR, and autonomous systems.

Recommended citation: Faiz Muhammad Chaudhry, Jarno Ralli, Jerome Leudet, Fahad Sohrab, Farhad Pakdaman, and Pierre Corbani. (2025). "DeepBrownConrady: Prediction of Camera Calibration and Distortion Parameters Using Deep Learning and Synthetic Data." IEEE Transactions on Automation Science and Engineering (T-ASE).
Download Paper