Hello neighbor: Accurate object retrieval with k-reciprocal nearest neighbors

By Danfeng Qin, Stephan Gammeter, Lukas Bossard, Till Quack, Luc Van Gool. Computer Vision and Pattern Recognition, 2011

344 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/2891033bc775ad5ca3737749f9de684d59335a64

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Ranking consistency for image matching and object retrieval (2014, methodology): “In contrast to [30], our method does not need to construct the whole network of dataset images.”
Fine-Tuning CNN Image Retrieval with No Human Annotation (2017, background): “Impressively, in this work we show that CNNs dominate the image search task by outperforming state-of-the-art methods that have reached a higher level of maturity by incorporating large visual codebooks [33], [34], spatial verification [33], [35] and query expansion [36], [37], [38].”
SIFT Meets CNN: A Decade Survey of Instance Retrieval (2016, background): “Other important extensions include “hello neighbor” based on reciprocal neighbors [116], QE with rank-based weighting [111], Hamming QE [89] (see Section 3.”
Pedestrian Alignment Network for Large-scale Person Re-Identification (2017, methodology): “Several re-ranking methods can be applied in person re-ID [Ye et al., 2015,Qin et al., 2011,Zhong et al., 2017].”
Particular object retrieval with integral max-pooling of CNN activations (2015, neutral): “d D Oxf5k Par6k Oxf105k Par106k Method Oxf5k Par6k Oxf105k Par106k J´egou & Zisserman (2014) 1024 56.0 - 50.2 - Chum et al. (2011) 82.7 80.5 76.7 71.0 J´egou & Zisserman (2014) 128 43.3 -35.3 Danfeng et al. (2011) 81.4 80.3 76.7 Babenko et al. (2014) 128 55.7 - 52.3 - Mikulik et al. (2013) 8”
End-to-End Learning of Deep Visual Representations for Image Retrieval (2016, methodology): “Finally, we would like to point out that our method uses a single universal model – the same for all test datasets –, contrary to, for instance, other methods of Danfeng et al (2011); Shen et al (2014); Tolias et al (2015) that perform some learning on the target datasets.”
Exploiting the Intrinsic Neighborhood Structure for Source-free Domain Adaptation (2021, background): “Reciprocal neighbors have been studied before in different contexts [14, 31, 60].”
A Pose-Sensitive Embedding for Person Re-identification with Expanded Cross Neighborhood Re-ranking (2017, methodology): “[24] formally uses the kreciprocal neighbors to compute ranking lists.”
Re-ranking Person Re-identification with k-Reciprocal Encoding (2017, methodology): “In literature, the �-reciprocal nearest neighbor [15, 35] is an effective solution to the above-mentioned problem, i.”
A Three-Layered Graph-Based Learning Approach for Remote Sensing Image Retrieval (2016, background): “For this reason, some relevant publications mainly focused on one reranked retrieval method [24] or text-based retrieval approach [25].”
Query-adaptive late fusion for image search and person re-identification (2015, neutral): “Methods Ours [40] [4] [35] [13] [24] [28] [12] [25] [11]”
Person Re-identification: Past, Present and Future (2016, methodology): “Leng et al. [168] use the idea of reciprocal k - nearest neighbors [169] to reﬁne the initial rank list based constructing images relations in the ofﬂine steps.”
Three things everyone should know to improve object retrieval (2012, methodology): “The precision (but not the recall) of the pipeline is improved by using spatial consistency to rerank a tf-idf short list [23, 28] or to rank all images with common visual words [8], or other (re)ranking variations [9, 27].”
Generalizing a Person Retrieval Model Hetero- and Homogeneously (2018, background): “Liu et al. [29] estimate labels with k-reciprocal nearest neighbors [31] and iteratively learn features for unsupervised video re-ID.”
Deep Image Retrieval: Learning Global Representations for Image Search (2016, methodology): “[10,14,59,63]) and perform a costly spatial verification (SV) at test time (e.”