Research map / Video / Video segmentation and tracking
360VOTS: Visual Object Tracking and Segmentation in Omnidirectional Videos
Yinzhe Xu, Huajian Huang, Yingshu Chen, Sai-Kit Yeung
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 360 tracking framework (ablation) | AiATrack · 360VOT BBox | S_dual: 0.405 -> 0.534 | Improved |
| add the 360 tracking framework (ablation) | SiamX · 360VOT BBox | S_dual: 0.302 -> 0.391 | Improved |
| add the 360 tracking framework (ablation) | XMem · 360VOS | J: 0.560 -> 0.656 | Improved |
| add the 360 tracking framework (ablation) | XMem* · 360VOS | J: 0.596 -> 0.676 | Improved |
| retrain on 360VOTS training set (ablation) | XMem · 360VOS | J: 0.560 -> 0.596 | Improved |
| fine-tune on 360VOTS training set (ablation) | LoRAT · 360VOT BBox | S_dual: 0.461 -> 0.495 | Improved |
| reduce SR Min from 90 to 60 deg (ablation) | XMem-360* · 360VOS | (J&F)_sphere: 0.739 -> 0.689 | Hurt |
| disable lost-target search region mechanism (Max Loss = 0 vs 4) (ablation) | XMem-360* · 360VOS | (J&F)_sphere: 0.701 -> 0.739 | Improved |
| vary SR Ratio from 2.0 to 1.2-3.6 (ablation) | XMem-360* · 360VOS | (J&F)_sphere: 0.739 -> 0.725 to 0.747 | No clear effect |
| apply spherical weighting to J near poles (ablation) | segmentation metric · 360VOS sample over-poles | J: 0.792 -> 0.575 | Mixed |
| convert VOS output to BFoV evaluation vs rBBox for AiATrack-360 | AiATrack-360 · 360VOT | S: 0.534 -> 0.362 | Hurt |
| use VOS tracker XMem-360 instead of AiATrack-360 on rBBox | tracking · 360VOT rBBox | P_angle: 0.535 -> 0.574 | 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.
- XMem++ — XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model (ECCV 2022) · better on 5 of 5 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
- 360VOT: A New Benchmark Dataset for Omnidirectional Visual Object Tracking (ICCV 2023 · similarity 0.94)
- Leader360V: A Large-scale, Real-world 360 Video Dataset for Multi-task Learning in Diverse Environment (NeurIPS 2025 · similarity 0.87)
- OmniLocalRF: Omnidirectional Local Radiance Fields from Dynamic Videos (CVPR 2024 · similarity 0.87)
- 360Loc: A Dataset and Benchmark for Omnidirectional Visual Localization with Cross-device Queries (CVPR 2024 · similarity 0.86)
- TAPVid-360: Tracking Any Point in 360 from Narrow Field of View Video (NeurIPS 2025 · similarity 0.86)
- OVT-B: A New Large-Scale Benchmark for Open-Vocabulary Multi-Object Tracking (NeurIPS 2024 · similarity 0.86)
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.
