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

Temporal action detection (TAD) with end-to-end training often suffers fromthe pain of huge demand for computing resources due to long video duration. Inthis work, we propose an efficient temporal action detector (ETAD) that cantrain directly from video frames with extremely low GPU memory consumption. Ourmain idea is to minimize and balance the heavy computation among features andgradients in each training iteration. We propose to sequentially forward thesnippet frame through the video encoder, and backward only a small necessaryportion of gradients to update the encoder. To further alleviate thecomputational redundancy in training, we propose to dynamically sample only asmall subset of proposals during training. Moreover, various samplingstrategies and ratios are studied for both the encoder and detector. ETADachieves state-of-the-art performance on TAD benchmarks with remarkableefficiency. On ActivityNet-1.3, training ETAD in 18 hours can reach 38.25%average mAP with only 1.3 GB memory consumption per video under end-to-endtraining. Our code will be publicly released.

ETAD: Training Action Detection End to End on a Laptop