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

3D Convolution Neural Networks (CNNs) have been widely applied to 3D sceneunderstanding, such as video analysis and volumetric image recognition.However, 3D networks can easily lead to over-parameterization which incursexpensive computation cost. In this paper, we propose Channel-wise AutomaticKErnel Shrinking (CAKES), to enable efficient 3D learning by shrinking standard3D convolutions into a set of economic operations e.g., 1D, 2D convolutions.Unlike previous methods, CAKES performs channel-wise kernel shrinkage, whichenjoys the following benefits: 1) enabling operations deployed in every layerto be heterogeneous, so that they can extract diverse and complementaryinformation to benefit the learning process; and 2) allowing for an efficientand flexible replacement design, which can be generalized to bothspatial-temporal and volumetric data. Further, we propose a new search spacebased on CAKES, so that the replacement configuration can be determinedautomatically for simplifying 3D networks. CAKES shows superior performance toother methods with similar model size, and it also achieves comparableperformance to state-of-the-art with much fewer parameters and computationalcosts on tasks including 3D medical imaging segmentation and video actionrecognition. Codes and models are available athttps://github.com/yucornetto/CAKES

CAKES: Channel-wise Automatic KErnel Shrinking for Efficient 3D Networks