In-shop Clothes Retrieval on In-shop Clothes Retrieval Dataset (test)
92.8Recall@1Intra-Batch Connections
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Intra-Batch ConnectionsEvaluation Protocol=Standard2021.02 | 92.8 | — | — | — | — | — | |
| Proxy-Anchorembedding size=512, larger input image=true2020.03 | 92.6 | 98.3 | 98.9 | — | 99.3 | — | |
| Intra-Batch ConnectionsEvaluation Protocol=Reality Check (Musgrave et al., 2020)2021.02 | 92.5 | — | — | — | — | — | |
| CCL*Backbone=ResNet-50, Embedding Size=512, margin (m)=nonzero, Input Size=Larger2023.08 | 92.3 | 98.5 | 99 | — | — | — | |
| CCLBackbone=ResNet-50, Embedding Size=512, margin (m)=0, Input Size=Larger2023.08 | 92.1 | 98.4 | 98.9 | — | — | — | |
| CGDBackbone=SE-ResNet-50, Dim=1536, Input Size=256x256, Configuration=GS2019.03 | 91.9 | 98.1 | 98.7 | 99 | 99.1 | 99.3 | |
| ProxyAnchor+Backbone=Inception, Embedding Size=5122023.08 | 91.9 | 98.1 | 98.7 | — | — | — | |
| MetrixBackbone=ResNet-50, Embedding Size=5122023.08 | 91.9 | 98.1 | 98.8 | — | — | — | |
| Proxy-Anchorembedding size=5122020.03 | 91.5 | 98.1 | 98.8 | — | 99.1 | — | |
| CGDBackbone=ResNet-50, Dim=1536, Input Size=256x256, Configuration=GS2019.03 | 90.9 | 98 | 98.7 | 99 | 99.1 | 99.2 | |
| Triplet-SCTBackbone=ResNet-50, Embedding Size=512, Input Size=Larger2023.08 | 90.9 | 97.5 | 98.1 | — | — | — | |
| FastAPBackbone=ResNet-50, Embedding Size=5122023.08 | 90.9 | 97.7 | 98.5 | — | — | — | |
| Proxy-Anchorembedding size=1282020.03 | 90.8 | 97.9 | 98.5 | — | 99 | — | |
| Cross-Entropyd=l2, Architecture=ResNet-502020.03 | 90.6 | 97.8 | 98.5 | 98.8 | 98.9 | 99 | |
| Cross-Entropyd=cos, Architecture=ResNet-502020.03 | 90.6 | 98 | 98.6 | 98.9 | 99.1 | 99.2 | |
| CEBackbone=ResNet-50, Embedding Size=20482023.08 | 90.6 | 98 | 98.6 | — | — | — | |
| Contra+HORDEembedding size=512, larger input image=true2020.03 | 90.4 | 97.8 | 98.4 | — | 98.9 | — | |
| HORDEBackbone=Inception, Embedding Size=512, Input Size=Larger2023.08 | 90.4 | 97.8 | 98.4 | — | — | — | |
| ProxyNCA++Backbone=ResNet-50, Embedding Size=5122023.08 | 90.4 | 98.1 | 98.8 | — | — | — | |
| BaselineBackbone=ResNet-50, Embedding Size=512, margin (m)=0, lambda=0, Input Size=Larger2023.08 | 90.2 | 98 | 98.7 | — | — | — | |
| XBMBackbone=Inception, Embedding Size=5122023.08 | 89.9 | 97.6 | 98.4 | — | — | — | |
| MSembedding size=5122020.03 | 89.7 | 97.9 | 98.5 | — | 99.1 | — | |
| Multi-Similarityd=cos, Architecture=BN-Inception2020.03 | 89.7 | 97.9 | 98.5 | 98.8 | 99.1 | 99.2 | |
| MSBackbone=Inception, Embedding Size=5122023.08 | 89.7 | 97.9 | 98.5 | — | — | — | |
| NormSoftmaxd=cos, Architecture=ResNet-502020.03 | 89.4 | 97.8 | 98.7 | 99 | — | — | |
| NSoftmaxBackbone=ResNet-50, Embedding Size=20482023.08 | 89.4 | 97.8 | 98.7 | — | — | — | |
| BFE+Backbone=ResNet-50, Dim=1536, Input Size=256x2562019.03 | 89.1 | 96.3 | 97.6 | 98.5 | 99.1 | — | |
| CGDBackbone=BN-Inception, Dim=128, Input Size=224x224, Configuration=GS2019.03 | 88.5 | 97.1 | 98 | 98.5 | 98.8 | 98.9 | |
| CGDBackbone=ResNet-50, Dim=128, Input Size=224x224, Configuration=GS2019.03 | 88.4 | 97.2 | 98.1 | 98.4 | 98.7 | 98.8 | |
| MIC+marginDim=1282019.09 | 88.2 | 97 | — | 98 | — | 98.8 | |
| BlackBoxBackbone=ResNet-50, Embedding Size=5122023.08 | 88.1 | 97 | 97.9 | — | — | — | |
| MSembedding size=1282020.03 | 88 | 97.2 | 98.1 | — | 98.7 | — | |
| EPSHNd=cos, Architecture=ResNet-502020.03 | 87.8 | 95.7 | 96.8 | 98 | — | — | |
| ABE-8Embedding Dimension=512, Number of attention-based learners=82018.04 | 87.3 | 96.7 | 97.9 | 98.2 | 98.5 | 98.7 | |
| ABE-8Embedding dimension=512, Number of learners (M)=82018.04 | 87.3 | 96.7 | 97.9 | 98.2 | 98.5 | 98.7 | |
| ABEembedding size=5122020.03 | 87.3 | 96.7 | 97.9 | — | 98.5 | — | |
| ABE-8Backbone=GoogleNet, Dim=512, Input Size=256x2562019.03 | 87.3 | 96.7 | 97.9 | 98.2 | 98.5 | 98.7 | |
| ABEd=cos, Architecture=GoogLeNet2020.03 | 87.3 | 96.7 | 97.9 | 98.2 | 98.5 | 98.7 | |
| ABE-4Embedding Dimension=512, Number of attention-based learners=42018.04 | 86.7 | 96.4 | 97.6 | 98 | 98.4 | 98.6 | |
| ABE-4Embedding dimension=512, Number of learners (M)=42018.04 | 86.7 | 96.4 | 97.6 | 98 | 98.4 | 98.6 | |
| CGDBackbone=BN-Inception, Dim=64, Input Size=224x224, Configuration=GS2019.03 | 86.6 | 96.3 | 97.4 | 97.9 | 98.2 | 98.4 | |
| CGDBackbone=ShuffleNet-v2, Dim=1536, Input Size=224x224, Configuration=GS2019.03 | 86.1 | 96.9 | 97.8 | 98.4 | 98.6 | 98.7 | |
| D&Cd=cos, Architecture=ResNet-502020.03 | 85.7 | 95.5 | 96.9 | 97.5 | — | — | |
| ABE-2Embedding Dimension=512, Number of attention-based learners=22018.04 | 85.2 | 96 | 97.2 | 97.8 | 98.2 | 98.4 | |
| ABE-2Embedding dimension=512, Number of learners (M)=22018.04 | 85.2 | 96 | 97.2 | 97.8 | 98.2 | 98.4 | |
| MarginDim=128, implementation=best re-implementation with ResNet502019.09 | 84.5 | 95.7 | — | 97.6 | — | 98.3 | |
| A-BIEREmbedding Dimension=5122018.04 | 83.1 | 95.1 | 96.9 | 97.5 | 97.8 | 98 | |
| A-BIEREmbedding dimension=5122018.04 | 83.1 | 95.1 | 96.9 | 97.5 | 97.8 | 98 | |
| A-BIERDim=5122019.09 | 83.1 | 95.1 | — | 97.5 | — | 98 | |
| A-BIERembedding size=5122020.03 | 83.1 | 95.1 | 96.9 | — | 97.8 | — | |
| HTLDim=5122019.09 | 80.9 | 94.3 | — | 97.2 | — | 97.8 | |
| HTLembedding size=1282020.03 | 80.9 | 94.3 | 95.8 | — | 97.4 | — | |
| HTLBackbone=BN-Inception, Dim=128, Input Size=224x2242019.03 | 80.9 | 94.3 | 95.8 | 97.2 | 97.4 | 97.8 | |
| HTLd=cos, Architecture=GoogLeNet2020.03 | 80.9 | 94.3 | 95.8 | 97.2 | 97.4 | 97.8 | |
| HTLBackbone=Inception, Embedding Size=5122023.08 | 80.9 | 94.3 | 95.8 | — | — | — | |
| HTG2019.09 | 80.3 | 93.9 | — | 96.6 | — | 97.1 | |
| DREMLDim=92162019.09 | 78.4 | 93.7 | — | 96.7 | — | — | |
| DREMLd=cos, Architecture=ResNet-182020.03 | 78.4 | 93.7 | 95.8 | 96.7 | — | — | |
| BIERDim=5122019.09 | 76.9 | 92.8 | — | 96.2 | — | 97.1 | |
| HDCEmbedding Dimension=3842018.04 | 62.1 | 84.9 | 89 | 91.2 | 92.3 | 93.1 | |
| HDCEmbedding dimension=3842018.04 | 62.1 | 84.9 | 89 | 91.2 | 92.3 | 93.1 | |
| HDCembedding size=3842020.03 | 62.1 | 84.9 | 89 | — | 92.3 | — | |
| HDCd=cos, Architecture=GoogLeNet2020.03 | 62.1 | 84.9 | 89 | 91.2 | 92.3 | 93.1 | |
| FasionNetEmbedding Dimension=40962018.04 | 53 | 73 | 76 | 77 | 79 | 80 | |
| FashionNetEmbedding dimension=40962018.04 | 53 | 73 | 76 | 77 | 79 | 80 | |
| FashionNetembedding size=40962020.03 | 53 | 73 | 76 | — | 79 | — | |
| FasionNet+PoseletsEmbedding Dimension=40962018.04 | 42 | 65 | 70 | 72 | 72 | 75 | |
| FashionNet+PoseletsEmbedding dimension=40962018.04 | 42 | 65 | 70 | 72 | 72 | 75 | |
| FasionNet+JointsEmbedding Dimension=40962018.04 | 41 | 64 | 68 | 71 | 73 | 73.5 | |
| FashionNet+JointsEmbedding dimension=40962018.04 | 41 | 64 | 68 | 71 | 73 | 73.5 | |
| GRNetBackbone=GoogleNet, Scales=7, Projection dimension=5122019.08 | 25.7 | — | 64.4 | — | — | 75 | |
| KPMBackbone=GoogleNet2019.08 | 21.3 | — | 54.1 | — | — | 65.2 | |
| DREMLBackbone=GoogleNet, Meta classes=192, Ensemble models=482019.08 | 18.6 | — | 51 | — | — | 59.1 | |
| VAM+ImgDropBackbone=GoogleNet2019.08 | 13.7 | — | 43.9 | — | — | 56.9 | |
| VAM+ProductBackbone=GoogleNet2019.08 | 13.4 | — | 43.6 | — | — | 56.7 | |
| VAM+NonsharedBackbone=GoogleNet2019.08 | 11.3 | — | 38.8 | — | — | 51.5 | |
| TripletBackbone=GoogleNet2019.08 | 10 | — | 37 | — | — | 49.9 | |
| FashionNet2019.08 | 7 | — | 18.8 | — | — | 22.8 |