Particular Object Retrieval With Integral Max-Pooling of CNN Activations

Abstract : Recently, image representation built upon Convolutional Neural Network (CNN) has been shown to provide effective descriptors for image search, outperform-ing pre-CNN features as short-vector representations. Yet such models are not compatible with geometry-aware re-ranking methods and still outperformed, on some particular object retrieval benchmarks, by traditional image search systems relying on precise descriptor matching, geometric re-ranking, or query expansion. This work revisits both retrieval stages, namely initial search and re-ranking, by employing the same primitive information derived from the CNN. We build compact feature vectors that encode several image regions without the need to feed multiple inputs to the network. Furthermore, we extend integral images to handle max-pooling on convolutional layer activations, allowing us to efficiently localize matching objects. The resulting bounding box is finally used for image re-ranking. As a result, this paper significantly improves existing CNN-based recognition pipeline: We report for the first time results competing with traditional methods on the challenging Oxford5k and Paris6k datasets.
Document type :
Conference papers
Complete list of metadatas

Cited literature [25 references]  Display  Hide  Download

https://hal.inria.fr/hal-01842218
Contributor : Teddy Furon <>
Submitted on : Wednesday, July 18, 2018 - 9:26:23 AM
Last modification on : Friday, September 13, 2019 - 9:48:07 AM
Long-term archiving on : Friday, October 19, 2018 - 4:20:56 PM

File

1511.05879.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01842218, version 1

Citation

Giorgos Tolias, Ronan Sicre, Hervé Jégou. Particular Object Retrieval With Integral Max-Pooling of CNN Activations. ICL 2016 - RInternational Conference on Learning Representations, May 2016, San Juan, Puerto Rico. pp.1-12. ⟨hal-01842218⟩

Share

Metrics

Record views

318

Files downloads

720