Particular object retrieval with integral max-pooling of CNN activations

18 Nov 2015  ·  Giorgos Tolias, Ronan Sicre, Hervé Jégou ·

Recently, image representation built upon Convolutional Neural Network (CNN) has been shown to provide effective descriptors for image search, outperforming 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.

PDF Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Image Retrieval Oxf105k R-MAC+R+QE MAP 73.2% # 7
Image Retrieval Oxf105k R-MAC MAP 61.6% # 8
Image Retrieval Par106k R-MAC+R+QE mAP 79.8% # 5
Image Retrieval Par106k R-MAC mAP 75.7% # 7
Image Retrieval Par6k R-MAC mAP 83.0% # 7
Image Retrieval Par6k R-MAC+R+QE mAP 86.5% # 4

Methods


No methods listed for this paper. Add relevant methods here