Research map / Video / Video segmentation and tracking
Spectrum-guided Multi-granularity Referring Video Object Segmentation
Bo Miao, Mohammed Bennamoun, Yongsheng Gao, Ajmal Mian
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 Conditional Patch Kernel (CPK) segmentation head (ablation) | SgMg · Ref-YouTube-VOS | J&F: 54.4 -> 55.8 | Improved |
| add Multi-granularity Segmentation Optimizer (MSO) (ablation) | SgMg · Ref-YouTube-VOS | J&F: 54.4 -> 57.7 | Improved |
| add Spectrum-guided Cross-modal Fusion (SCF) (ablation) | SgMg · Ref-YouTube-VOS | J&F: 54.4 -> 55.8 | Improved |
| segment encoded features directly (segment-and-optimize) instead of decode-and-segment (ablation) | ReferFormer baseline · Ref-YouTube-VOS | J&F: 56.0 -> 57.7 | Improved |
| add MSO on top of decode-and-segment pipeline (ablation) | ReferFormer baseline · Ref-YouTube-VOS | J&F: 56.0 -> 56.4 | No clear effect |
| use spectral convolution instead of spatial convolution in SCF (ablation) | SCF · Ref-YouTube-VOS | J&F: 57.6 -> 58.9 | Improved |
| use spectral convolution instead of linear layer in SCF (ablation) | SCF · Ref-YouTube-VOS | J&F: 57.9 -> 58.9 | Improved |
| share features across multiple expressions (Fast SgMg multi-object R-VOS) | SgMg · Ref-YouTube-VOS | J&F / FPS: J&F 56.0 -> 54.2; FPS 50 -> 185 (vs ReferFormer) | Mixed |
| pretrain on RefCOCO/+/g and use Video-Swin-B backbone | SgMg · Ref-YouTube-VOS | J&F: 58.9 -> 65.7 | Improved |
Closest papers by meaning
- Referred by Multi-Modality: A Unified Temporal Transformer for Video Object Segmentation (AAAI 2024 · similarity 0.89)
- Few-Shot Referring Video Single- and Multi-Object Segmentation Via Cross-Modal Affinity with Instance Sequence Matching (IJCV 2025 · similarity 0.89)
- Language As Queries for Referring Video Object Segmentation (CVPR 2022 · similarity 0.89)
- QST-SAM: Leveraging Cross-modal Instructions for Few-shot Referring Video Object Segmentation (ECCV 2026 · similarity 0.88)
- Learning Cross-Modal Affinity for Referring Video Object Segmentation Targeting Limited Samples (ICCV 2023 · similarity 0.88)
- Multi-Granularity Video Object Segmentation (AAAI 2025 · 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.
