VIFT Official PyTorch Implementation of Causal Transformer for Fusion and Pose Estimation in Deep Visual Inertial Odometry [ECCVW2024]