Table 1 Datasets created using nuScenes and MOTFront data with the number of samples (pairs of percept-anchor) in the training, validation and test splits.

From: SAILOR: perceptual anchoring for robotic cognitive architectures