List your company
ResearchFormerly WizWand

Deep Metric Learning on SOP (test)

81.4Recall@1Intra-Batch Connections
68.19271.62175.0578.479Mar 24, 2020Jun 5, 2020Aug 18, 2020Oct 31, 2020Jan 12, 2021Mar 27, 2021Jun 9, 2021
Updated 8mo ago

Evaluation results

MethodLinks
Intra-Batch ConnectionsEvaluation Protocol=Standard2021.0281.4—————
PA + MetrixBackbone=ResNet-50, Embedding size=512, Metrix variant=feature mixup2021.0681.3———91.796.9
ProxyNCA++ + MetrixBackbone=ResNet-50, Embedding size=512, Metrix variant=feature mixup2021.0681.3———92.797.1
Intra-Batch ConnectionsEvaluation Protocol=Reality Check (Musgrave et al., 2020)2021.0281.1—————
MS + MetrixBackbone=ResNet-50, Embedding size=512, Metrix variant=feature mixup2021.0681———9297.2
ProxyNCA++Backbone=ResNet-50, Embedding size=512, source=reported2021.0680.7———9296.7
RankDim=15362020.0379.891.396.390.4——
Margin + MSDFBaseline Objective=Margin Loss, beta=1.2, S2SD Variant=Multiscale Self-Distillation with Features, Backbone=ResNet-502020.0979.63——90.7——
Margin + MSDBaseline Objective=Margin Loss, beta=1.2, S2SD Variant=Multiscale Self-Distillation, Backbone=ResNet-502020.0979.26——90.6——
Margin + DSDBaseline Objective=Margin Loss, beta=1.2, S2SD Variant=Dual Self-Distillation, Backbone=ResNet-502020.0979.05——90.52——
Margin + MSDFABaseline Objective=Margin Loss, beta=1.2, S2SD Variant=Multiscale Self-Distillation with Features and Affinities, Backbone=ResNet-502020.0978.82——90.49——
Multisimilarity + MSDFBaseline Objective=Multisimilarity Loss, S2SD Variant=Multiscale Self-Distillation with Features, Backbone=ResNet-502020.0978.59——90.09——
R-Margin + MSDFBaseline Objective=Relaxed Margin Loss, beta=0.6, S2SD Variant=Multiscale Self-Distillation with Features, Backbone=ResNet-502020.0978.57——90.58——
MarginBaseline Objective=Margin Loss, beta=1.2, Backbone=ResNet-502020.0978.43——90.4——
Multisimilarity + MSDBaseline Objective=Multisimilarity Loss, S2SD Variant=Multiscale Self-Distillation, Backbone=ResNet-502020.0978.42——90.09——
SOFT-TRIPLEDim=5122020.0378.390.395.992——
Multisimilarity + DSDBaseline Objective=Multisimilarity Loss, S2SD Variant=Dual Self-Distillation, Backbone=ResNet-502020.0978.23——90.08——
Multisimilarity + MSDFABaseline Objective=Multisimilarity Loss, S2SD Variant=Multiscale Self-Distillation with Features and Affinities, Backbone=ResNet-502020.0978.07——89.88——
R-Margin + MSDBaseline Objective=Relaxed Margin Loss, beta=0.6, S2SD Variant=Multiscale Self-Distillation, Backbone=ResNet-502020.0978——90.47——
R-Margin + MSDFABaseline Objective=Relaxed Margin Loss, beta=0.6, S2SD Variant=Multiscale Self-Distillation with Features and Affinities, Backbone=ResNet-502020.0978——90.41——
MultisimilarityBaseline Objective=Multisimilarity Loss, Backbone=ResNet-502020.0977.99——90——
R-Margin + DSDBaseline Objective=Relaxed Margin Loss, beta=0.6, S2SD Variant=Dual Self-Distillation, Backbone=ResNet-502020.0977.86——90.5——
R-MarginBaseline Objective=Relaxed Margin Loss, beta=0.6, Backbone=ResNet-502020.0977.58——90.42——
MICDim=1282020.0377.289.495.690——
Ours (Margin + PADS)Dim=1282020.0376.58995.489.9——
ABEDim=5122020.0376.388.494.8———
D&CDim=1282020.0375.988.494.990.2——
HTLDim=5122020.0374.888.394.8———
A-BIERDim=5122020.0374.286.994———
MarginDim=1282020.0372.786.293.890.8——
DVMLDim=5122020.0370.285.293.890.8——
HDMLDim=5122020.0368.783.292.489.3——