GENESIS-V2: Inferring Unordered Object Representations without Iterative Refinement

Benchmark Model Rank Results
unsupervised-object-segmentation-on-shelf-tote-trainingGENESIS-V2#1ARI: 0.55
unsupervised-object-segmentation-on-shelf-tote-trainingMONET-G#2ARI: 0.11
unsupervised-object-segmentation-on-shelf-tote-trainingGENESIS#3ARI: 0.04
unsupervised-object-segmentation-on-shelf-tote-trainingSlotAttention#4ARI: 0.03
unsupervised-object-segmentation-on-objectsroomGENESIS-V2#2ARI-FG: 0.84
unsupervised-object-segmentation-on-objectsroomSlotAttention#3ARI-FG: 0.79
unsupervised-object-segmentation-on-objectsroomGENESIS#4ARI-FG: 0.63
unsupervised-object-segmentation-on-objectsroomMONET-G#5ARI-FG: 0.54
unsupervised-object-segmentation-on-shapestacksGENESIS-V2#2ARI-FG: 0.81
unsupervised-object-segmentation-on-shapestacksSlotAttention#3ARI-FG: 0.76
unsupervised-object-segmentation-on-shapestacksGENESIS#4ARI-FG: 0.70
unsupervised-object-segmentation-on-shapestacksMONET-G#5ARI-FG: 0.70