HyperAI超神経

プラットフォーム

Visual Place Recognition On Mapillary Val

評価指標

Recall@1

Recall@10

Recall@5

評価結果

このベンチマークにおける各モデルのパフォーマンス結果

				Paper Title
Pair-VPR-p	95.4	97.7	97.3	Pair-VPR: Place-Aware Pre-training and Contrastive Pair Classification for Visual Place Recognition with Vision Transformers
EffoVPR	92.8	97.4	97.2	EffoVPR: Effective Foundation Model Utilization for Visual Place Recognition
Pair-VPR-s	93.7	97.3	97.2	Pair-VPR: Place-Aware Pre-training and Contrastive Pair Classification for Visual Place Recognition with Vision Transformers
SelaVPR	90.8	97.2	96.4	Towards Seamless Adaptation of Pre-trained Models for Visual Place Recognition
BoQ	93.8	97	96.8	BoQ: A Place is Worth a Bag of Learnable Queries
DINOv2 SALAD	92.2	97	96.4	Optimal Transport Aggregation for Visual Place Recognition
BoQ (ResNet-50)	91.2	96.1	95.3	BoQ: A Place is Worth a Bag of Learnable Queries
MixVPR	88.2	94.3	93.1	MixVPR: Feature Mixing for Visual Place Recognition
CosPlace (ResNet-101 2048-D)	86.7	93.4	92.1	Rethinking Visual Geo-localization for Large-Scale Applications
ResNeXt GCL + PCA	80.9	92.6	90.7	Generalized Contrastive Optimization of Siamese Networks for Place Recognition
Conv-AP	83.4	92.3	90.5	GSV-Cities: Toward Appropriate Supervised Visual Place Recognition
CosPlace	-	91.8	89.9	Rethinking Visual Geo-localization for Large-Scale Applications
NetVLAD (with GPM)	82.0	91.4	90.4	Global Proxy-based Hard Mining for Visual Place Recognition
ResNeXt GCL	75.5	88.5	86.1	Generalized Contrastive Optimization of Siamese Networks for Place Recognition
Patch-NetVLAD	79.5	87.7	86.2	Patch-NetVLAD: Multi-Scale Fusion of Locally-Global Descriptors for Place Recognition

0 of 15 row(s) selected.

Visual Place Recognition On Mapillary Val | SOTA | HyperAI超神経