Abstract:Conventional multi-view re-ranking methods usually perform asymmetrical matching between the region of interest (ROI) in the query image and the whole target image for similarity computation. Due to the inconsistency in the visual appearance, this practice tends to degrade the retrieval accuracy particularly when the image ROI, which is usually interpreted as the image objectness, accounts for a smaller region in the image. Since Privileged Information (PI), which can be viewed as the image prior, is able to characterize well the image objectness, we are aiming at leveraging PI for further improving the performance of multi-view re-ranking in this paper. Towards this end, we propose a discriminative multi-view re-ranking approach in which both the original global image visual contents and the local auxiliary PI features are simultaneously integrated into a unified training framework for generating the latent subspaces with sufficient discriminating power. For the on-the-fly re-ranking, since the multi-view PI features are unavailable, we only project the original multi-view image representations onto the latent subspace, and thus the re-ranking can be achieved by computing and sorting the distances from the multi-view embeddings to the separating hyperplane. Extensive experimental evaluations on the two public benchmarks, Oxford5k and Paris6k, reveal that our approach provides further performance boost for accurate image re-ranking, whilst the comparative study demonstrates the advantage of our method against other multi-view re-ranking methods.

Listwise View Ranking for Image Cropping

Discriminate Clearer to Rank Better: Image Cropping by Amplifying View-Wise Differences.

An End-to-End Neural Network for Image Cropping by Learning Composition from Aesthetic Photos

Image Cropping with Spatial-Aware Feature and Rank Consistency

Saliency Aware Image Cropping with Latent Region Pair.

Image Cropping with Composition and Saliency Aware Aesthetic Score Map

Automatic Image Cropping for Visual Aesthetic Enhancement Using Deep Neural Networks and Cascaded Regression

Emotion Attention-Aware Collaborative Deep Reinforcement Learning for Image Cropping

ROMIR: Robust Multi-View Image Re-Ranking

Image Search Re-Ranking with an Improved VisualRank Algorithm and Multi-Layer DCNN Features.

A Deep Network Solution for Attention and Aesthetics Aware Photo Cropping

Cropper: Vision-Language Model for Image Cropping through In-Context Learning

A2-RL: Aesthetics Aware Reinforcement Learning for Image Cropping

Context-Aware Candidates for Image Cropping

Repurposing existing deep networks for caption and aesthetic-guided image cropping

Deep Cropping via Attention Box Prediction and Aesthetics Assessment

Learning a Deep Listwise Context Model for Ranking Refinement

A composition-oriented aesthetic view recommendation network supervised by the simplified golden ratio theory

Joint Probability Distribution Regression for Image Cropping

Discriminative Multi-view Privileged Information Learning for Image Re-ranking

From Document to Image: Learning a Scalable Ranking Model for Content Based Image Retrieval