AI Paper Atlas

Research map / Video / Video segmentation and tracking

Diversifying Spatial-Temporal Perception for Video Domain Generalization

Kun-Yu Lin, Jia-Run Du, Yipeng Gao, Jiaming Zhou, Wei-Shi Zheng

NeurIPS 2023 · Video › Video segmentation and tracking · approach: Self-supervised video representation learning · 6 citations (OpenAlex)

Read the paperCheck 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
Add spatial grouping to the backbone (ablation)STDN · UCF-HMDB UCF→HMDBaccuracy: 52.7 -> 54.9Improved
Add spatial grouping to the backbone (ablation)STDN · UCF-HMDB HMDB→UCFaccuracy: 71.9 -> 73.9Improved
Add temporal dependency modeling after spatial grouping (ablation)Backbone + SGM · UCF-HMDB UCF→HMDBaccuracy: 54.9 -> 56.7Improved
Add temporal dependency modeling after spatial grouping (ablation)Backbone + SGM · UCF-HMDB HMDB→UCFaccuracy: 73.9 -> 75.3Improved
Use full spatial-temporal relation modeling rather than the preceding TRM variant (ablation)Backbone + SGM · UCF-HMDB UCF→HMDBaccuracy: 54.9 -> 58.3Improved
Use full spatial-temporal relation modeling rather than the preceding TRM variant (ablation)Backbone + SGM · UCF-HMDB HMDB→UCFaccuracy: 73.9 -> 76.2Improved
Add video-adapted MixStyle (ablation)Backbone + STRM · UCF-HMDB UCF→HMDBaccuracy: 58.3 -> 59.3Improved
Add video-adapted MixStyle (ablation)Backbone + STRM · UCF-HMDB HMDB→UCFaccuracy: 76.2 -> 76.6Improved
Add the full STDN feature aggregation and complete design (ablation)Backbone + MixStyle · UCF-HMDB UCF→HMDBaccuracy: 59.3 -> 60.2Improved
Add the full STDN feature aggregation and complete design (ablation)Backbone + MixStyle · UCF-HMDB HMDB→UCFaccuracy: 76.6 -> 77.1Improved
Add relation discrimination loss to temporal relation modeling (ablation)STDN temporal relation module · UCF-HMDB UCF→HMDBnormalized MSE diversityImproved
Use STDN instead of VideoDGUCF-HMDB · HMDB→UCFaccuracy: 74.9 -> 77.1Improved

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