English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

The human visual system can effortlessly recognize an object under differentextrinsic factors such as lighting, object poses, and background, yet currentcomputer vision systems often struggle with these variations. An important stepto understanding and improving artificial vision systems is to measure imagesimilarity purely based on intrinsic object properties that define objectidentity. This problem has been studied in the computer vision literature asre-identification, though mostly restricted to specific object categories suchas people and cars. We propose to extend it to general object categories,exploring an image similarity metric based on object intrinsics. To benchmarksuch measurements, we collect the Common paired objects Under differenTExtrinsics (CUTE) dataset of $18,000$ images of $180$ objects under differentextrinsic factors such as lighting, poses, and imaging conditions. Whileexisting methods such as LPIPS and CLIP scores do not measure object intrinsicswell, we find that combining deep features learned from contrastiveself-supervised learning with foreground filtering is a simple yet effectiveapproach to approximating the similarity. We conduct an extensive survey ofpre-trained features and foreground extraction methods to arrive at a strongbaseline that best measures intrinsic object-centric image similarity amongcurrent methods. Finally, we demonstrate that our approach can aid indownstream applications such as acting as an analog for human subjects andimproving generalizable re-identification. Please see our project website athttps://s-tian.github.io/projects/cute/ for visualizations of the data anddemos of our metric.

Are These the Same Apple? Comparing Images Based on Object Intrinsics