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

Optical flow, which computes the apparent motion from a pair of video frames,is a critical tool for scene motion estimation. Correlation volume is thecentral component of optical flow computational neural models. It estimates thepairwise matching costs between cross-frame features, and is then used todecode optical flow. However, traditional correlation volume is frequentlynoisy, outlier-prone, and sensitive to motion blur. We observe that, althoughthe recent RAFT algorithm also adopts the traditional correlation volume, itsadditional context encoder provides semantically representative features to theflow decoder, implicitly compensating for the deficiency of the correlationvolume. However, the benefits of this context encoder has been barely discussedor exploited. In this paper, we first investigate the functionality of RAFT'scontext encoder, then propose a new Context Guided Correlation Volume (CGCV)via gating and lifting schemes. CGCV can be universally integrated withRAFT-based flow computation methods for enhanced performance, especiallyeffective in the presence of motion blur, de-focus blur and atmosphericeffects. By incorporating the proposed CGCV with previous Global MotionAggregation (GMA) method, at a minor cost of 0.5% extra parameters, the rank ofGMA is lifted by 23 places on KITTI 2015 Leader Board, and 3 places on SintelLeader Board. Moreover, at a similar model size, our correlation volumeachieves competitive or superior performance to state of the art peersupervised models that employ Transformers or Graph Reasoning, as verified byextensive experiments.

CGCV:Context Guided Correlation Volume for Optical Flow Neural Networks