Matches in SemOpenAlex for { <https://semopenalex.org/work/W4312770831> ?p ?o ?g. }
- W4312770831 abstract "Face identification (FI) is ubiquitous and drives many high-stake decisions made by the law enforcement. A common FI approach compares two images by taking the cosine similarity between their image embeddings. Yet, such approach suffers from poor out-of-distribution (OOD) generalization to new types of images (e.g., when a query face is masked, cropped or rotated) not included in the training set or the gallery. Here, we propose a re-ranking approach that compares two faces using the Earth Mover's Distance on the deep, spatial features of image patches. Our extra comparison stage explicitly examines image similarity at a fine-grained level (e.g., eyes to eyes) and is more robust to OOD perturbations and occlusions than traditional FI. Interestingly, without finetuning feature extractors, our method consistently improves the accuracy on all tested OOD queries: masked, cropped, rotated, and adversarial while obtaining similar results on in-distribution images." @default.
- W4312770831 created "2023-01-05" @default.
- W4312770831 creator A5007723039 @default.
- W4312770831 creator A5071289093 @default.
- W4312770831 date "2022-06-01" @default.
- W4312770831 modified "2023-10-16" @default.
- W4312770831 title "DeepFace-EMD: Re-ranking Using Patch-wise Earth Mover's Distance Improves Out-Of-Distribution Face Identification" @default.
- W4312770831 cites W1884784862 @default.
- W4312770831 cites W1932198206 @default.
- W4312770831 cites W1950843348 @default.
- W4312770831 cites W1984926403 @default.
- W4312770831 cites W2003907441 @default.
- W4312770831 cites W2085400714 @default.
- W4312770831 cites W2101149304 @default.
- W4312770831 cites W2108840547 @default.
- W4312770831 cites W2109685970 @default.
- W4312770831 cites W2129812935 @default.
- W4312770831 cites W2143668817 @default.
- W4312770831 cites W2146090483 @default.
- W4312770831 cites W2171235719 @default.
- W4312770831 cites W2404498690 @default.
- W4312770831 cites W2535873859 @default.
- W4312770831 cites W2565575913 @default.
- W4312770831 cites W2593886340 @default.
- W4312770831 cites W2663800299 @default.
- W4312770831 cites W2739522917 @default.
- W4312770831 cites W2778195589 @default.
- W4312770831 cites W2790299973 @default.
- W4312770831 cites W2805258252 @default.
- W4312770831 cites W2941841895 @default.
- W4312770831 cites W2962898354 @default.
- W4312770831 cites W2963202462 @default.
- W4312770831 cites W2963322158 @default.
- W4312770831 cites W2963466847 @default.
- W4312770831 cites W2963839617 @default.
- W4312770831 cites W2963953476 @default.
- W4312770831 cites W2964023530 @default.
- W4312770831 cites W2964350391 @default.
- W4312770831 cites W2969985801 @default.
- W4312770831 cites W2981408674 @default.
- W4312770831 cites W3000996870 @default.
- W4312770831 cites W3006027522 @default.
- W4312770831 cites W3012255272 @default.
- W4312770831 cites W3035470592 @default.
- W4312770831 cites W3035613124 @default.
- W4312770831 cites W3085402119 @default.
- W4312770831 cites W3099206234 @default.
- W4312770831 cites W3101998545 @default.
- W4312770831 cites W3109684201 @default.
- W4312770831 cites W3110144845 @default.
- W4312770831 cites W3143373604 @default.
- W4312770831 cites W3161655641 @default.
- W4312770831 cites W3186860856 @default.
- W4312770831 cites W3188008942 @default.
- W4312770831 cites W78725893 @default.
- W4312770831 doi "https://doi.org/10.1109/cvpr52688.2022.01962" @default.
- W4312770831 hasPublicationYear "2022" @default.
- W4312770831 type Work @default.
- W4312770831 citedByCount "5" @default.
- W4312770831 countsByYear W43127708312022 @default.
- W4312770831 countsByYear W43127708312023 @default.
- W4312770831 crossrefType "proceedings-article" @default.
- W4312770831 hasAuthorship W4312770831A5007723039 @default.
- W4312770831 hasAuthorship W4312770831A5071289093 @default.
- W4312770831 hasBestOaLocation W43127708312 @default.
- W4312770831 hasConcept C103278499 @default.
- W4312770831 hasConcept C115961682 @default.
- W4312770831 hasConcept C116834253 @default.
- W4312770831 hasConcept C134306372 @default.
- W4312770831 hasConcept C138885662 @default.
- W4312770831 hasConcept C144024400 @default.
- W4312770831 hasConcept C153180895 @default.
- W4312770831 hasConcept C154945302 @default.
- W4312770831 hasConcept C177148314 @default.
- W4312770831 hasConcept C177264268 @default.
- W4312770831 hasConcept C189430467 @default.
- W4312770831 hasConcept C199360897 @default.
- W4312770831 hasConcept C2776401178 @default.
- W4312770831 hasConcept C2779304628 @default.
- W4312770831 hasConcept C2780762811 @default.
- W4312770831 hasConcept C31510193 @default.
- W4312770831 hasConcept C31972630 @default.
- W4312770831 hasConcept C33923547 @default.
- W4312770831 hasConcept C36289849 @default.
- W4312770831 hasConcept C41008148 @default.
- W4312770831 hasConcept C41895202 @default.
- W4312770831 hasConcept C52622490 @default.
- W4312770831 hasConcept C59822182 @default.
- W4312770831 hasConcept C82668687 @default.
- W4312770831 hasConcept C86803240 @default.
- W4312770831 hasConceptScore W4312770831C103278499 @default.
- W4312770831 hasConceptScore W4312770831C115961682 @default.
- W4312770831 hasConceptScore W4312770831C116834253 @default.
- W4312770831 hasConceptScore W4312770831C134306372 @default.
- W4312770831 hasConceptScore W4312770831C138885662 @default.
- W4312770831 hasConceptScore W4312770831C144024400 @default.
- W4312770831 hasConceptScore W4312770831C153180895 @default.
- W4312770831 hasConceptScore W4312770831C154945302 @default.
- W4312770831 hasConceptScore W4312770831C177148314 @default.
- W4312770831 hasConceptScore W4312770831C177264268 @default.