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

Inspired by frame-based methods, state-of-the-art event-based optical flownetworks rely on the explicit construction of correlation volumes, which areexpensive to compute and store, at the same time prohibiting them fromestimating high-resolution flow. We observe that the spatiotemporallycontinuous traces of events provide a natural search direction for seekingpixel correspondences, obviating the need to rely on gradients of explicitcorrelation volumes as such search directions. We introduce IDNet (IterativeDeblurring Network), a lightweight yet high-performing event-based optical flownetwork directly estimating flow from event traces without using correlationvolumes. We further propose two iterative update schemes: "ID" which iteratesover the same batch of events, and "TID" which iterates over time withstreaming events in an online fashion. Benchmark results show the "ID" schemeoutperforms previous state-of-the-art by 9% of EPE with 52% fewer parametersand 77% savings in memory footprint, while the "TID" scheme is even moreefficient promising 80% of compute savings and 18 times less latency at thecost of only 6% of performance drop.

Rethinking Event-based Optical Flow: Iterative Deblurring as an Alternative to Correlation Volumes