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

Autonomous systems not only need to understand their current environment, butshould also be able to predict future actions conditioned on past states, forinstance based on captured camera frames. However, existing models mainly focuson forecasting future video frames for short time-horizons, hence being oflimited use for long-term action planning. We propose Multi-Scale HierarchicalPrediction (MSPred), a novel video prediction model able to simultaneouslyforecast future possible outcomes of different levels of granularity atdifferent spatio-temporal scales. By combining spatial and temporaldownsampling, MSPred efficiently predicts abstract representations such ashuman poses or locations over long time horizons, while still maintaining acompetitive performance for video frame prediction. In our experiments, wedemonstrate that MSPred accurately predicts future video frames as well ashigh-level representations (e.g. keypoints or semantics) on bin-picking andaction recognition datasets, while consistently outperforming popularapproaches for future frame prediction. Furthermore, we ablate differentmodules and design choices in MSPred, experimentally validating that combiningfeatures of different spatial and temporal granularity leads to a superiorperformance. Code and models to reproduce our experiments can be found inhttps://github.com/AIS-Bonn/MSPred.

MSPred: Video Prediction at Multiple Spatio-Temporal Scales with Hierarchical Recurrent Networks