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

Batch Normalization's (BN) unique property of depending on other samples in abatch is known to cause problems in several tasks, including sequentialmodeling. Yet, BN-related issues are hardly studied for long videounderstanding, despite the ubiquitous use of BN in CNNs for feature extraction.Especially in surgical workflow analysis, where the lack of pretrained featureextractors has lead to complex, multi-stage training pipelines, limitedawareness of BN issues may have hidden the benefits of training CNNs andtemporal models end to end. In this paper, we %present and analyze known aswell as novel pitfalls of BN in video learning, including issues specific toonline tasks such as a 'cheating' effect in anticipation. We observe that BN'sproperties create major obstacles for end-to-end learning. However, usingBN-free backbones, even simple CNN-LSTMs beat state of the art in two surgicaltasks by utilizing adequate end-to-end training strategies which maximizetemporal context. We conclude that awareness of BN's pitfalls is crucial foreffective end-to-end learning in surgical tasks. By reproducing results onnatural-video datasets, we hope our insights will benefit other areas of videolearning as well. Code: \url{https://gitlab.com/nct_tso_public/pitfalls_bn}.

On the Pitfalls of Batch Normalization for End-to-End Video Learning: A Study on Surgical Workflow Analysis