Google Landmarks Dataset v2 -- A Large-Scale Benchmark for Instance-Level Recognition and Retrieval

3 Apr 2020Tobias WeyandAndre AraujoBingyi CaoJack Sim

While image retrieval and instance recognition techniques are progressing rapidly, there is a need for challenging datasets to accurately measure their performance -- while posing novel challenges that are relevant for practical applications. We introduce the Google Landmarks Dataset v2 (GLDv2), a new benchmark for large-scale, fine-grained instance recognition and image retrieval in the domain of human-made and natural landmarks... (read more)

PDF Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Landmark Recognition Google Landmarks Dataset v2 (recognition, testing) DELG global+SP microAP 56.35 # 1
Landmark Recognition Google Landmarks Dataset v2 (recognition, validation) DELG global+SP microAP 55.01 # 1
Image Retrieval Google Landmarks Dataset v2 (retrieval, testing) ResNet101+ArcFace GLDv2-train-clean [email protected] 24.15 # 1
Image Retrieval Google Landmarks Dataset v2 (retrieval, validation) ResNet101+ArcFace GLDv2-train-clean [email protected] 22.2 # 1
Image Retrieval ROxford (Hard) ResNet101+ArcFace GLDv2-teain-clean mAP 51.6 # 1
Image Retrieval ROxford (Medium) ResNet101+ArcFace GLDv2-teain-clean mAP 74.2 # 1
Image Retrieval RParis (Hard) ResNet101+ArcFace GLDv2-teain-clean mAP 70.3 # 1
Image Retrieval RParis (Medium) ResNet101+ArcFace GLDv2-teain-clean mAP 84.9 # 1

Methods used in the Paper