Predict images without imu data/ training model without imu data

[viettiepx2]

Thank you for your work on eWaSR. I'm very impressed with the results presented in your paper.
I'm currently trying to use eWaSR to predict on images without IMU data, and I'm encountering some difficulties. It seems the model is expecting IMU input, and I'm unsure how to adapt it for image-only prediction. I'm also having trouble training the model without IMU data. (Error: Line 69 in model.py: features['imu_mask'] = x['imu_mask'].float().unsqueeze(1))
Could you please provide some guidance on how to modify the model and training process for image-only input? Any help would be greatly appreciated.
Thank you for your time and consideration.