Research map / Video / Video segmentation and tracking
Panoramic Vision Transformer for Saliency Detection in 360° Videos
Heeseung Yun, Sehun Lee, Gunhee Kim
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 global context projection MLP_G (ablation) | PAVER · Wild360 test | CC: 0.376 -> 0.616 | Improved |
| decouple global and local context encoders (ablation) | PAVER · Wild360 test | CC: 0.492 -> 0.616 | Improved |
| use 360-aware geometry modules instead of NFoV ones (ablation) | PAVER · Wild360 test | CC: 0.549 -> 0.616 | Improved |
| use spatiotemporal local fusion instead of simple MLP (ablation) | PAVER · Wild360 test | CC: 0.561 -> 0.616 | Improved |
| train with feature-level consistency instead of score-level (ablation) | PAVER · Wild360 test | CC: 0.575 -> 0.616 | Improved |
| add temporal saliency score to local+spatial (ablation) | PAVER · Wild360 test | CC: 0.585 -> 0.616 | Improved |
| add local saliency score to temporal+spatial (ablation) | PAVER · Wild360 test | CC: 0.557 -> 0.616 | Improved |
| compute saliency from temporal score alone (ablation) | PAVER · Wild360 test | CC: 0.616 -> 0.276 | Hurt |
| replace ViT-B/16 backbone with TimeSformer (ablation) | PAVER · Wild360 test | CC: 0.616 -> 0.465 | Hurt |
| replace ViT-B/16 with smaller patch DINO B/8 vs B/16 (ablation) | PAVER backbone · Wild360 test | CC: 0.524 -> 0.563 | Improved |
| replace ViT-B/16 with ViT-Ti/16 (ablation) | PAVER · Wild360 test | CC: 0.616 -> 0.540 | Hurt |
| increase input resolution from 224 to 384 (ablation) | PAVER · Wild360 test | CC: 0.616 -> 0.567 | Hurt |
Closest papers by meaning
- Visual Saliency Transformer (ICCV 2021 · similarity 0.87)
- PanoVOS: Bridging Non-panoramic and Panoramic Views with Transformer for Video Segmentation (ECCV 2024 · similarity 0.87)
- VST++: Efficient and Stronger Visual Saliency Transformer (TPAMI 2024 · similarity 0.86)
- Spherical Vision Transformers for Audio-Visual Saliency Prediction in 360$^{\circ }$∘ Videos (TPAMI 2025 · similarity 0.86)
- Saliency Detection in 360° Videos (ECCV 2018 · similarity 0.86)
- PanoDiT: Panoramic Videos Generation with Diffusion Transformer (AAAI 2025 · similarity 0.84)
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.
