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 Tools annual

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 monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Video prediction methods generally consume substantial computing resources intraining and deployment, among which keypoint-based approaches show promisingimprovement in efficiency by simplifying dense image prediction to lightkeypoint prediction. However, keypoint locations are often modeled only ascontinuous coordinates, so noise from semantically insignificant deviations invideos easily disrupt learning stability, leading to inaccurate keypointmodeling. In this paper, we design a new grid keypoint learning framework,aiming at a robust and explainable intermediate keypoint representation forlong-term efficient video prediction. We have two major technicalcontributions. First, we detect keypoints by jumping among candidate locationsin our raised grid space and formulate a condensation loss to encouragemeaningful keypoints with strong representative capability. Second, weintroduce a 2D binary map to represent the detected grid keypoints and thensuggest propagating keypoint locations with stochasticity by selecting entriesin the discrete grid space, thus preserving the spatial structure of keypointsin the longterm horizon for better future frame generation. Extensiveexperiments verify that our method outperforms the state-ofthe-art stochasticvideo prediction methods while saves more than 98% of computing resources. Wealso demonstrate our method on a robotic-assisted surgery dataset withpromising results. Our code is available athttps://github.com/xjgaocs/Grid-Keypoint-Learning.

Accurate Grid Keypoint Learning for Efficient Video Prediction