Exploiting local features from deep networks for image retrieval

Ng, Joe Yue-Hei; Yang, Fan; Davis, Larry S.

doi:10.1109/cvprw.2015.7301272

Cited by 283 publications

(215 citation statements)

References 23 publications

Supporting

Mentioning

212

Contrasting

Unclassified

Order By: Relevance

“…Features are extracted in a single pass using CNN pre-trained on some large-scale datasets like ImageNet [23]. Compact Encoding/pooling techniques are used [9], [10].…”

Section: Categorization Methodologymentioning

confidence: 99%

“…These filters produces n = 96 heat maps of size 27 × 27 (after max pooling). Each pixel in the maps has a receptive field of 19 × 19 and records the response of the image w.r.t the corresponding filter [9], [10], [134]. The column feature is therefore of size 1 × 1 × 96 (Fig.…”

Section: Feature Extractionmentioning

confidence: 99%

“…Many recent retrieval methods [9], [10], [134] focus on local descriptors in the intermediate layers. In these methods, lower-level convolutional filters (kernels) are used to detect local visual patterns.…”

Section: Feature Extractionmentioning

confidence: 99%

“…Babenko et al [8] were the first to fine-tune a CNN model for generic instance retrieval. Both [9], [10] employ the column features from pre-trained CNN models, and [10] inspires later state-of-the-art methods. These milestones are the representative works of the categorization scheme in this survey.…”

Section: 1 Approximate K-meansmentioning

confidence: 99%

“…In 2012, Krizhevsky et al [6] with the AlexNet achieved the state-of-the-art recognition accuracy in ILSRVC 2012, exceeding previous best results by a large margin. Since then, research focus has begun to transfer to deep learning based methods [7], [8], [9], [10], especially the convolutional neural network (CNN).…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

SIFT Meets CNN: A Decade Survey of Instance Retrieval

Zheng

Yang

Tian

2018

IEEE Trans. Pattern Anal. Mach. Intell.

656

319

View full text Add to dashboard Cite

Abstract-In the early days, content-based image retrieval (CBIR) was studied with global features. Since 2003, image retrieval based on local descriptors (de facto SIFT) has been extensively studied for over a decade due to the advantage of SIFT in dealing with image transformations. Recently, image representations based on the convolutional neural network (CNN) have attracted increasing interest in the community and demonstrated impressive performance. Given this time of rapid evolution, this article provides a comprehensive survey of instance retrieval over the last decade. Two broad categories, SIFT-based and CNN-based methods, are presented. For the former, according to the codebook size, we organize the literature into using large/medium-sized/small codebooks. For the latter, we discuss three lines of methods, i.e., using pre-trained or fine-tuned CNN models, and hybrid methods. The first two perform a single-pass of an image to the network, while the last category employs a patch-based feature extraction scheme. This survey presents milestones in modern instance retrieval, reviews a broad selection of previous works in different categories, and provides insights on the connection between SIFT and CNN-based methods. After analyzing and comparing retrieval performance of different categories on several datasets, we discuss promising directions towards generic and specialized instance retrieval.

show abstract

“…Features are extracted in a single pass using CNN pre-trained on some large-scale datasets like ImageNet [23]. Compact Encoding/pooling techniques are used [9], [10].…”

Section: Categorization Methodologymentioning

confidence: 99%

Section: Feature Extractionmentioning

confidence: 99%

Section: Feature Extractionmentioning

confidence: 99%

Section: 1 Approximate K-meansmentioning

confidence: 99%