Metric Learning on Stanford Online Products

Metric: R@1 (higher is better)

LeaderboardDataset

Loading chart...

Results

Hide augmentations

Sort:

#	Model↕	R@1▼	Augmentations	Paper	Date↕	Code
1	Unicom+ViT-L@336px	91.2	Yes	Unicom: Universal and Compact Representation Lea...	2023-04-12	Code
2	STIR	88.3	No	STIR: Siamese Transformer for Image Retrieval Po...	2023-04-26	Code
3	Recall@k Surrogate Loss (ViT-B/16)	88	No	Recall@k Surrogate Loss with Large Batches and S...	2021-08-25	Code
4	ViT-Triplet	86.5	No	STIR: Siamese Transformer for Image Retrieval Po...	2023-04-26	Code
5	ROADMAP (DeiT-S)	86	No	Robust and Decomposable Average Precision for Im...	2021-10-01	Code
6	Hyp-ViT	85.9	Yes	Hyperbolic Vision Transformers: Combining Improv...	2022-03-21	Code
7	Hyp-DINO	85.1	No	Hyperbolic Vision Transformers: Combining Improv...	2022-03-21	Code
8	Recall@k Surrogate Loss (ViT-B/32)	85.1	No	Recall@k Surrogate Loss with Large Batches and S...	2021-08-25	Code
9	ROADMAP (ResNet-50)	83.1	No	Robust and Decomposable Average Precision for Im...	2021-10-01	Code
10	CCL (ResNet-50)	83.1	No	Center Contrastive Loss for Metric Learning	2023-08-01	-
11	Recall@k Surrogate Loss (ResNet-50)	82.7	No	Recall@k Surrogate Loss with Large Batches and S...	2021-08-25	Code
12	Gradient Surgery	82.3	No	Dissecting the impact of different loss function...	2022-01-27	-
13	HAPPIER_F	81.8	Yes	Hierarchical Average Precision Training for Pert...	2022-07-05	Code
14	SCT(512)	81.6	No	Hard negative examples are hard, but useful	2020-07-24	Code
15	ResNet-50 + Metrix	81.3	Yes	It Takes Two to Tango: Mixup for Deep Metric Lea...	2021-06-09	Code
16	ResNet50 + Language	81.3	Yes	Integrating Language Guidance into Vision-based ...	2022-03-16	Code
17	NED	81.2	No	Calibrated neighborhood aware confidence measure...	2020-06-08	-
18	ResNet-50 + Cross-Entropy	81.1	Yes	A unifying mutual information view of metric lea...	2020-03-19	Code
19	ResNet50 + S2SD	81	Yes	S2SD: Simultaneous Similarity-based Self-Distill...	2020-09-17	Code
20	HAPPIER	81	Yes	Hierarchical Average Precision Training for Pert...	2022-07-05	Code
21	ResNet-50 + ProxyNCA++	80.7	Yes	ProxyNCA++: Revisiting and Revitalizing Proxy Ne...	2020-04-02	Code
22	ResNet50 + NIR	80.7	Yes	Non-isotropy Regularization for Proxy-based Deep...	2022-03-16	Code
23	MS + DAS (K=8)	80.59	No	DAS: Densely-Anchored Sampling for Deep Metric L...	2022-07-30	Code
24	BN-Inception + Proxy-Anchor	80.3	No	Proxy Anchor Loss for Deep Metric Learning	2020-03-31	Code
25	ResNet50 + DiVA	79.6	Yes	DiVA: Diverse Visual Feature Aggregation for Dee...	2020-04-28	Code
26	ResNet50 + AVSL	79.6	Yes	Attributable Visual Similarity Learning	2022-03-28	Code
27	Margin + DIML	79.26	No	Towards Interpretable Deep Metric Learning with ...	2021-08-12	Code
28	Circle Loss	78.3	No	Circle Loss: A Unified Perspective of Pair Simil...	2020-02-25	Code
29	EPSHN(512)	78.3	No	Improved Embeddings with Easy Positive Triplet M...	2019-04-08	Code
30	QB-Norm+RDML	78.1	No	Cross Modal Retrieval with Querybank Normalisation	2021-12-23	Code
31	ResNet50 (128) + MIC	77.2	Yes	MIC: Mining Interclass Characteristics for Impro...	2019-09-25	Code
32	ResNet50 (128) + PADS	76.5	Yes	PADS: Policy-Adapted Sampling for Visual Similar...	2020-03-24	Code
33	Group Loss	75.7	No	The Group Loss for Deep Metric Learning	2019-12-01	Code

#1Unicom+ViT-L@336pxSOTA
91.2
R@1· Augmentations· 2023-04-12
Unicom: Universal and Compact Representation Learning for Image Retrieval Code
#2STIR
88.3
R@1· 2023-04-26
STIR: Siamese Transformer for Image Retrieval Postprocessing Code
#3Recall@k Surrogate Loss (ViT-B/16)SOTA
88
R@1· 2021-08-25
Recall@k Surrogate Loss with Large Batches and Similarity Mixup Code
#4ViT-Triplet
86.5
R@1· 2023-04-26
STIR: Siamese Transformer for Image Retrieval Postprocessing Code
#5ROADMAP (DeiT-S)
86
R@1· 2021-10-01
Robust and Decomposable Average Precision for Image Retrieval Code
#6Hyp-ViT
85.9
R@1· Augmentations· 2022-03-21
Hyperbolic Vision Transformers: Combining Improvements in Metric Learning Code
#7Hyp-DINO
85.1
R@1· 2022-03-21
Hyperbolic Vision Transformers: Combining Improvements in Metric Learning Code
#8Recall@k Surrogate Loss (ViT-B/32)
85.1
R@1· 2021-08-25
Recall@k Surrogate Loss with Large Batches and Similarity Mixup Code
#9ROADMAP (ResNet-50)
83.1
R@1· 2021-10-01
Robust and Decomposable Average Precision for Image Retrieval Code
#10CCL (ResNet-50)
83.1
R@1· 2023-08-01
Center Contrastive Loss for Metric Learning
#11Recall@k Surrogate Loss (ResNet-50)
82.7
R@1· 2021-08-25
Recall@k Surrogate Loss with Large Batches and Similarity Mixup Code
#12Gradient Surgery
82.3
R@1· 2022-01-27
Dissecting the impact of different loss functions with gradient surgery
#13HAPPIER_F
81.8
R@1· Augmentations· 2022-07-05
Hierarchical Average Precision Training for Pertinent Image Retrieval Code
#14SCT(512)SOTA
81.6
R@1· 2020-07-24
Hard negative examples are hard, but useful Code
#15ResNet-50 + Metrix
81.3
R@1· Augmentations· 2021-06-09
It Takes Two to Tango: Mixup for Deep Metric Learning Code
#16ResNet50 + Language
81.3
R@1· Augmentations· 2022-03-16
Integrating Language Guidance into Vision-based Deep Metric Learning Code
#17NEDSOTA
81.2
R@1· 2020-06-08
Calibrated neighborhood aware confidence measure for deep metric learning
#18ResNet-50 + Cross-EntropySOTA
81.1
R@1· Augmentations· 2020-03-19
A unifying mutual information view of metric learning: cross-entropy vs. pairwise losses Code
#19ResNet50 + S2SD
81
R@1· Augmentations· 2020-09-17
S2SD: Simultaneous Similarity-based Self-Distillation for Deep Metric Learning Code
#20HAPPIER
81
R@1· Augmentations· 2022-07-05
Hierarchical Average Precision Training for Pertinent Image Retrieval Code
#21ResNet-50 + ProxyNCA++
80.7
R@1· Augmentations· 2020-04-02
ProxyNCA++: Revisiting and Revitalizing Proxy Neighborhood Component Analysis Code
#22ResNet50 + NIR
80.7
R@1· Augmentations· 2022-03-16
Non-isotropy Regularization for Proxy-based Deep Metric Learning Code
#23MS + DAS (K=8)
80.59
R@1· 2022-07-30
DAS: Densely-Anchored Sampling for Deep Metric Learning Code
#24BN-Inception + Proxy-Anchor
80.3
R@1· 2020-03-31
Proxy Anchor Loss for Deep Metric Learning Code
#25ResNet50 + DiVA
79.6
R@1· Augmentations· 2020-04-28
DiVA: Diverse Visual Feature Aggregation for Deep Metric Learning Code
#26ResNet50 + AVSL
79.6
R@1· Augmentations· 2022-03-28
Attributable Visual Similarity Learning Code
#27Margin + DIML
79.26
R@1· 2021-08-12
Towards Interpretable Deep Metric Learning with Structural Matching Code
#28Circle Loss
78.3
R@1· 2020-02-25
Circle Loss: A Unified Perspective of Pair Similarity Optimization Code
#29EPSHN(512)SOTA
78.3
R@1· 2019-04-08
Improved Embeddings with Easy Positive Triplet Mining Code
#30QB-Norm+RDML
78.1
R@1· 2021-12-23
Cross Modal Retrieval with Querybank Normalisation Code
#31ResNet50 (128) + MIC
77.2
R@1· Augmentations· 2019-09-25
MIC: Mining Interclass Characteristics for Improved Metric Learning Code
#32ResNet50 (128) + PADS
76.5
R@1· Augmentations· 2020-03-24
PADS: Policy-Adapted Sampling for Visual Similarity Learning Code
#33Group Loss
75.7
R@1· 2019-12-01
The Group Loss for Deep Metric Learning Code