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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Event analysis in untrimmed videos has attracted increasing attention due tothe application of cutting-edge techniques such as CNN. As a well studiedproperty for CNN-based models, the receptive field is a measurement formeasuring the spatial range covered by a single feature response, which iscrucial in improving the image categorization accuracy. In video domain, videoevent semantics are actually described by complex interaction among differentconcepts, while their behaviors vary drastically from one video to another,leading to the difficulty in concept-based analytics for accurate eventcategorization. To model the concept behavior, we study temporal conceptreceptive field of concept-based event representation, which encodes thetemporal occurrence pattern of different mid-level concepts. Accordingly, weintroduce temporal dynamic convolution (TDC) to give stronger flexibility toconcept-based event analytics. TDC can adjust the temporal concept receptivefield size dynamically according to different inputs. Notably, a set ofcoefficients are learned to fuse the results of multiple convolutions withdifferent kernel widths that provide various temporal concept receptive fieldsizes. Different coefficients can generate appropriate and accurate temporalconcept receptive field size according to input videos and highlight crucialconcepts. Based on TDC, we propose the temporal dynamic concept modelingnetwork (TDCMN) to learn an accurate and complete concept representation forefficient untrimmed video analysis. Experiment results on FCVID and ActivityNetshow that TDCMN demonstrates adaptive event recognition ability conditioned ondifferent inputs, and improve the event recognition performance ofConcept-based methods by a large margin. Code is available athttps://github.com/qzhb/TDCMN.

Modeling Temporal Concept Receptive Field Dynamically for Untrimmed Video Analysis