Image Retrieval On Roxford Hard

评估指标

mAP

评测结果

各个模型在此基准测试上的表现结果

		Paper Title	Repository
SuperGlobal	80.2	Global Features are All You Need for Image Retrieval and Reranking
AMES	80	AMES: Asymmetric and Memory-Efficient Similarity Estimation for Instance-level Retrieval
Hypergraph propagation+community selection	73	Hypergraph Propagation and Community Selection for Objects Retrieval	-
Token	66.57	Learning Token-based Representation for Image Retrieval
DELG+ α QE reranking+ RRT reranking	64	Instance-level Image Retrieval using Reranking Transformers
FIRe	61.2	Learning Super-Features for Image Retrieval
HOW	56.9	Learning and aggregating deep local descriptors for instance-level recognition
ResNet101+ArcFace GLDv2-train-clean	51.6	Google Landmarks Dataset v2 -- A Large-Scale Benchmark for Instance-Level Recognition and Retrieval
DELF–HQE+SP	50.3	Large-Scale Image Retrieval with Attentive Deep Local Features
HesAff–rSIFT–HQE+SP	49.7	Revisiting Oxford and Paris: Large-Scale Image Retrieval Benchmarking
DELF–ASMK*+SP	43.1	Large-Scale Image Retrieval with Attentive Deep Local Features
HesAff–rSIFT–HQE	41.3	Revisiting Oxford and Paris: Large-Scale Image Retrieval Benchmarking
R–GeM	38.5	Fine-tuning CNN Image Retrieval with No Human Annotation
HesAff–rSIFT–ASMK*+SP	36.7	Revisiting Oxford and Paris: Large-Scale Image Retrieval Benchmarking
HesAff–rSIFT–ASMK*	36.4	Revisiting Oxford and Paris: Large-Scale Image Retrieval Benchmarking
HesAff–rSIFT–SMK*+SP	35.8	Revisiting Oxford and Paris: Large-Scale Image Retrieval Benchmarking
HesAff–rSIFT–SMK*	35.4	Revisiting Oxford and Paris: Large-Scale Image Retrieval Benchmarking
R–R-MAC	32.4	Particular object retrieval with integral max-pooling of CNN activations
Dino	24.3	Emerging Properties in Self-Supervised Vision Transformers
R – [O] –MAC	18.0	Particular object retrieval with integral max-pooling of CNN activations

0 of 23 row(s) selected.