Research map / Video / Video segmentation and tracking
PanoSAM2: Lightweight Distortion- and Memory-aware Adaptions of SAM2 for 360 Video Object Segmentation
Dingwen Xiao, WEIMING ZHANG, Shiqi Wen, Addison Wang
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 Pano-Aware Decoder (ablation) | SAM2 (Hiera-T) · 360VOTS test (Hiera-T) | J&F: 59.4 -> 62.7 | Improved |
| add Long-Short Memory Module (ablation) | SAM2 (Hiera-T) · 360VOTS test (Hiera-T) | J&F: 59.4 -> 61.6 | Improved |
| add distortion-guided mask loss (ablation) | SAM2 (Hiera-T) · 360VOTS test (Hiera-T) | J&F: 59.4 -> 60.1 | Improved |
| combine PA decoder, LSMM and mask loss (ablation) | SAM2 (Hiera-T) · 360VOTS test (Hiera-T) | J&F: 59.4 -> 64.3 | Improved |
| add PC Block to PA decoder (ablation) | PA decoder · 360VOTS test (Hiera-T) | J&F: 62.2 -> 63.8 | Improved |
| add previous-frame mask to PA decoder (ablation) | PA decoder · 360VOTS test (Hiera-T) | J&F: 62.2 -> 62.9 | Improved |
| combine PC Block and previous-frame mask (ablation) | PA decoder · 360VOTS test (Hiera-T) | J&F: 62.2 -> 64.3 | Improved |
| add long-term memory with L=2 (ablation) | PanoSAM2 · 360VOTS test (Hiera-T) | J&F: 63.4 -> 64.3 | Improved |
| increase long-term memory size from 2 to 3 (ablation) | PanoSAM2 · 360VOTS test (Hiera-T) | J&F: 64.3 -> 63.7 | Hurt |
| add occlusion loss weight lambda_occ=0.1 (ablation) | PanoSAM2 · 360VOTS test (Hiera-T) | J&F: 63.1 -> 64.3 | Improved |
| use weight range w_min=0.5, w_max=2.0 instead of uniform 1.0 (ablation) | distortion-guided loss · 360VOTS test (Hiera-T) | J&F: 63.8 -> 64.3 | Improved |
| widen weight range to w_min=0.25, w_max=4.0 (ablation) | distortion-guided loss · 360VOTS test (Hiera-T) | J&F: 64.3 -> 63.6 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- SAM2 — SAM 2: Segment Anything in Images and Videos (ICLR 2025) · better on 8 of 8 shared results
- SAM2Long — SAM2Long: Enhancing SAM 2 for Long Video Segmentation with a Training-Free Memory Tree (ICCV 2025) · better on 4 of 4 shared results
- XMem — XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model (ECCV 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
- SAM-I2V: Upgrading SAM to Support Promptable Video Segmentation with Less than 0.2% Training Cost (CVPR 2025 · similarity 0.89)
- SAM 2: Segment Anything in Images and Videos (ICLR 2025 · similarity 0.88)
- Leader360V: A Large-scale, Real-world 360 Video Dataset for Multi-task Learning in Diverse Environment (NeurIPS 2025 · similarity 0.88)
- LiVOS: Light Video Object Segmentation with Gated Linear Matching (CVPR 2025 · similarity 0.86)
- CamSAM2: Segment Anything Accurately in Camouflaged Videos (NeurIPS 2025 · similarity 0.85)
- SAMWISE: Infusing Wisdom in SAM2 for Text-Driven Video Segmentation (CVPR 2025 · similarity 0.85)
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.
