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

Recently, 3D convolutional networks yield good performance in actionrecognition. However, optical flow stream is still needed to ensure betterperformance, the cost of which is very high. In this paper, we propose a fastbut effective way to extract motion features from videos utilizing residualframes as the input data in 3D ConvNets. By replacing traditional stacked RGBframes with residual ones, 20.5% and 12.5% points improvements over top-1accuracy can be achieved on the UCF101 and HMDB51 datasets when trained fromscratch. Because residual frames contain little information of objectappearance, we further use a 2D convolutional network to extract appearancefeatures and combine them with the results from residual frames to form atwo-path solution. In three benchmark datasets, our two-path solution achievedbetter or comparable performances than those using additional optical flowmethods, especially outperformed the state-of-the-art models on Mini-kineticsdataset. Further analysis indicates that better motion features can beextracted using residual frames with 3D ConvNets, and our residual-frame-inputpath is a good supplement for existing RGB-frame-input models.

Rethinking Motion Representation: Residual Frames with 3D ConvNets for Better Action Recognition