Are you combining the 3 dataset: Nuscenes, Nuscenes-Occupancy, Lyft-Level5, or did you train the model and evaluate on 3 datasets seperately?
Can I preprocess only on Nuscenes and train/evaluate only on Nuscenes? Or if you combined 3 datasets, is it possibile to train and evaluate on the combination of Nuscenes and Nuscenes-Occupancy only?
Just curious, from the loss function, the model learns from GT occupancy constraints, in this case, assuming DriveOccWorld model used additional GT info from Nuscenes-Occupancy, should we consider it as additional data, did you re-train other E2E models such as UniAD, VAD with more GT data from Nuscenes-Occupancy?
Are you combining the 3 dataset: Nuscenes, Nuscenes-Occupancy, Lyft-Level5, or did you train the model and evaluate on 3 datasets seperately?
Can I preprocess only on Nuscenes and train/evaluate only on Nuscenes? Or if you combined 3 datasets, is it possibile to train and evaluate on the combination of Nuscenes and Nuscenes-Occupancy only?
Just curious, from the loss function, the model learns from GT occupancy constraints, in this case, assuming DriveOccWorld model used additional GT info from Nuscenes-Occupancy, should we consider it as additional data, did you re-train other E2E models such as UniAD, VAD with more GT data from Nuscenes-Occupancy?