Research map / Video / Video segmentation and tracking
Dual-Path Adaptation From Image to Video Transformers
Jungin Park, Jiyoung Lee, Kwanghoon Sohn
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add the temporal adaptation path (ablation) | DualPath · Something-Something-v2 | R@1: 53.7 -> 69.3 | Improved |
| add the spatial adaptation path (ablation) | DualPath · Something-Something-v2 | R@1: 55.1 -> 69.3 | Improved |
| increase spatial frames from T_S=8 to T_S=12 (ablation) | DualPath · Something-Something-v2 | R@1: 69.3 -> 69.6 | Improved |
| increase spatial frames from T_S=8 to T_S=16 (ablation) | DualPath · Something-Something-v2 | R@1: 69.3 -> 69.8 | Improved |
| set grid scaling factors to w=h=2 instead of w=h=4 (ablation) | DualPath · Something-Something-v2 | R@1: 71.8 -> 66.4 | Hurt |
| set grid scaling factors to w=h=8 instead of w=h=4 (ablation) | DualPath · Something-Something-v2 | R@1: 71.8 -> 61.5 | Hurt |
| increase grid resolution from 224×224 to 448×448 (ablation) | DualPath · Something-Something-v2 | R@1: 69.3 -> 70.5 | Improved |
| increase grid resolution from 224×224 to 896×896 (ablation) | DualPath · Something-Something-v2 | R@1: 69.3 -> 71.6 | Improved |
| use 48 frames instead of 16 at 224×224 grid resolution (ablation) | DualPath · Something-Something-v2 | R@1: 69.3 -> 71.2 | Improved |
| use DualPath with ViT-L/14 rather than X-CLIP-L/14 | Kinetics-400 · Kinetics-400 | R@1: 87.1 -> 87.7 | Improved |
| use DualPath rather than ST-Adapter with ViT-B/16 | ViT-B/16 · Something-Something-v2 | R@1: 69.5 -> 70.3 | Improved |
| use DualPath rather than VideoPrompt | ViT-B/16 · HMDB51 | accuracy: 66.4 -> 75.6 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ST-Adapter — ST-Adapter: Parameter-Efficient Image-to-Video Transfer Learning (NeurIPS 2022) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- AIM: Adapting Image Models for Efficient Video Action Recognition (ICLR 2023 · similarity 0.88)
- Video Transformers: A Survey (TPAMI 2023 · similarity 0.88)
- D2ST-Adapter: Disentangled-and-Deformable Spatio-Temporal Adapter for Few-shot Action Recognition (ICCV 2025 · similarity 0.87)
- Optimizing Factorized Encoder Models: Time and Memory Reduction for Scalable and Efficient Action Recognition (ECCV 2024 · similarity 0.87)
- Adapting Short-Term Transformers for Action Detection in Untrimmed Videos (CVPR 2024 · similarity 0.87)
- ViViT: A Video Vision Transformer (ICCV 2021 · similarity 0.87)
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.
