Towards Content-based Pixel Retrieval in Revisited Oxford and Paris

TL;DR

This paper presents the first pixel retrieval benchmarks, PROxford and PRParis, designed to evaluate fine-grained, pixel-level object retrieval, and demonstrates that current state-of-the-art methods struggle with this challenging task.

Contribution

Introduction of two new pixel retrieval benchmarks based on existing datasets, with detailed annotations and extensive evaluation of current methods.

Findings

Pixel retrieval is more challenging than existing image retrieval tasks.

Current SOTA methods perform poorly on the new pixel retrieval benchmarks.

The datasets enable further research in fine-grained, pixel-level content-based retrieval.

Abstract

This paper introduces the first two pixel retrieval benchmarks. Pixel retrieval is segmented instance retrieval. Like semantic segmentation extends classification to the pixel level, pixel retrieval is an extension of image retrieval and offers information about which pixels are related to the query object. In addition to retrieving images for the given query, it helps users quickly identify the query object in true positive images and exclude false positive images by denoting the correlated pixels. Our user study results show pixel-level annotation can significantly improve the user experience. Compared with semantic and instance segmentation, pixel retrieval requires a fine-grained recognition capability for variable-granularity targets. To this end, we propose pixel retrieval benchmarks named PROxford and PRParis, which are based on the widely used image retrieval datasets, ROxford and RParis. Three professional annotators label 5,942 images with two rounds of double-checking and refinement. Furthermore, we conduct extensive experiments and analysis on the SOTA methods in image search, image matching, detection, segmentation, and dense matching using our pixel retrieval benchmarks. Results show that the pixel retrieval task is challenging to these approaches and distinctive from existing problems, suggesting that further research can advance the content-based pixel-retrieval and thus user search experience. The datasets can be downloaded from \href{https://github.com/anguoyuan/Pixel_retrieval-Segmented_instance_retrieval}{this link}.