DropMAE: Masked Autoencoders with Spatial-Attention Dropout for Tracking Tasks

Benchmark Model Rank Results
visual-object-tracking-on-lasot-extDropTrack#16AUC: 52.7Precision: 60.2
visual-object-tracking-on-tnl2kDropTrack#19AUC: 56.9precision: 57.9
visual-object-tracking-on-got-10kDropMAE#27Average Overlap: 75.9Success Rate 0.5: 86.8Success Rate 0.75: 72
visual-object-tracking-on-lasotDropTrack#28AUC: 71.8Normalized Precision: 81.8Precision: 78.1
visual-object-tracking-on-trackingnetDropTrack#45Normalized Precision: 88.9AUC: 0.841