RefVOS: A Closer Look at Referring Expressions for Video Object Segmentation

Benchmark Model Rank Results
referring-expression-segmentation-on-a2dRefVOS#16IoU overall: 0.599IoU mean: 0.599Precision@0.5: 0.495
referring-expression-segmentation-on-davisRefVOS#13J&F 1st frame: 44.5J&F Full video: 45.1
referring-expression-segmentation-on-refcocoRefVOS with BERT + MLM loss#27Overall IoU: 59.45
referring-expression-segmentation-on-refcocoRefVOS with BERT Pre-train#29Overall IoU: 58.65
referring-expression-segmentation-on-refcoco-3RefVOS with BERT + MLM loss#26Overall IoU: 44.71
referring-expression-segmentation-on-refcoco-4RefVOS with BERT + MLM Loss#23Overall IoU: 49.73
referring-expression-segmentation-on-refcoco-5RefVOS with BERT + MLM loss#23Overall IoU: 36.17