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

Video object detection needs to solve feature degradation situations thatrarely happen in the image domain. One solution is to use the temporalinformation and fuse the features from the neighboring frames. WithTransformerbased object detectors getting a better performance on the imagedomain tasks, recent works began to extend those methods to video objectdetection. However, those existing Transformer-based video object detectorsstill follow the same pipeline as those used for classical object detectors,like enhancing the object feature representations by aggregation. In this work,we take a different perspective on video object detection. In detail, weimprove the qualities of queries for the Transformer-based models byaggregation. To achieve this goal, we first propose a vanilla query aggregationmodule that weighted averages the queries according to the features of theneighboring frames. Then, we extend the vanilla module to a more practicalversion, which generates and aggregates queries according to the features ofthe input frames. Extensive experimental results validate the effectiveness ofour proposed methods: On the challenging ImageNet VID benchmark, whenintegrated with our proposed modules, the current state-of-the-artTransformer-based object detectors can be improved by more than 2.4% on mAP and4.2% on AP50.

FAQ: Feature Aggregated Queries for Transformer-based Video Object Detectors