AI Paper Atlas

Research map / Video / Video segmentation and tracking

Learning Semantic-Aware Dynamics for Video Prediction

Xinzhu Bei, Yanchao Yang, Stefano Soatto

CVPR 2021 · Video › Video segmentation and tracking · approach: Stochastic video prediction · 31 citations (OpenAlex)

Check your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
model class-specific semantic dynamics and use semantic-informed dis-occlusion synthesisvideo prediction · Cityscapes t+5MS-SSIM: 75.68 -> 83.51Improved
model class-specific semantic dynamics and use semantic-informed dis-occlusion synthesisvideo prediction · Cityscapes t+5LPIPS: 16.50 -> 14.93Improved
model class-specific semantic dynamics and use semantic-informed dis-occlusion synthesisvideo prediction · KITTI RawMS-SSIM: 79.28 -> 83.06; 67.65 -> 72.44; 60.77 -> 64.72Improved
model class-specific semantic dynamics and use semantic-informed dis-occlusion synthesisvideo prediction · KITTI Raw t+3LPIPS: 24.61 -> 24.58Improved
model class-specific semantic dynamics and use semantic-informed dis-occlusion synthesisvideo prediction · KITTI Raw t+5LPIPS: 30.49 -> 31.16Hurt
model class-specific semantic dynamics and use semantic-informed dis-occlusion synthesisvideo prediction · KITTI Flow next-framePSNR: 22.3 -> 24.47Improved
model class-specific semantic dynamics and use semantic-informed dis-occlusion synthesisvideo prediction · KITTI Flow next-frameSSIM: 69.6 -> 71.1Improved
model class-specific semantic dynamics and use semantic-informed dis-occlusion synthesisvideo prediction · KITTI Flow next-frameLPIPS: 11.4 -> 10.9Improved
merge static classes to use 9 classes rather than 20 (ablation)SADM · Cityscapes video prediction t+5MS-SSIM: 79.14 -> 83.51Improved
merge static classes to use 9 classes rather than 20 (ablation)SADM · Cityscapes video prediction t+5LPIPS: 17.55 -> 14.93Improved
replace class-specific dynamics with a single-class model using concatenated masks and flow (ablation)SADM · video predictionmotion accuracy near semantic boundariesHurt
swap car and road semantic encoders (ablation)SADM · qualitative motion predictionclass-consistent motionHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy