Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Seek to Segment: Active Perception for Panoramic Referring Segmentation
Song Tang, Shuming Hu, Xincheng Shuai, Henghui Ding, Yu-Gang Jiang
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 EgoSphere spatial memory (ablation) | zero-shot Qwen3-VL-8B agent · APRS | SR / SPL: SR 41.6 -> 56.5; SPL 0.14 -> 0.26 | Improved |
| add supervised fine-tuning on expert trajectories (ablation) | EgoSphere agent · APRS | SR: 56.5 -> 70.3 | Improved |
| add GRPO reinforcement learning (ablation) | SFT + EgoSphere agent · APRS | SPL / AS: SPL 0.40 -> 0.57; AS 6.2 -> 4.8 | Improved |
| add GRPO reinforcement learning (ablation) | SFT + EgoSphere agent · APRS | SR: 70.3 -> 75.4 | Improved |
| add textual log memory (ablation) | zero-shot agent · APRS | SR: 41.6 -> 50.3 | Improved |
| add 5-frame visual buffer (ablation) | zero-shot agent · APRS | SR: 41.6 -> 46.8 | Improved |
| add hybrid text log and visual buffer (ablation) | zero-shot agent · APRS | SR: 46.8 -> 43.6 | Hurt |
| use EgoSphere instead of textual log memory (ablation) | zero-shot agent · APRS | SR / SPL: SR 50.3 -> 56.5; SPL 0.21 -> 0.26 | Improved |
| use active VLM agent with EgoSphere vs heuristic scanning (VisionReasoner) | PanoSeeker · APRS | SPL: 0.26 -> 0.57 | Improved |
| use PanoSeeker over GPT-5.2 text-log agent | GPT-5.2 · APRS | SPL: 0.36 -> 0.57 | Improved |
| use active exploration over static single-pass RIS (VisionReasoner) | VisionReasoner · APRS | SR: 66.2 -> 75.4 | Improved |
| replace full-panorama input with fixed heuristic scanning | LISA · APRS | SR: 64.1 -> 42.3 | Hurt |
Closest papers by meaning
- Mask Grounding for Referring Image Segmentation (CVPR 2024 · similarity 0.86)
- SaFiRe: Saccade-Fixation Reiteration with Mamba for Referring Image Segmentation (NeurIPS 2025 · similarity 0.86)
- LQMFormer: Language-aware Query Mask Transformer for Referring Image Segmentation (CVPR 2024 · similarity 0.86)
- Think Before You Segment: An Object-aware Reasoning Agent for Referring Audio-Visual Segmentation (AAAI 2026 · similarity 0.86)
- SetCon: Towards Open-Ended Referring Segmentation via Set-Level Concept Prediction (NeurIPS 2026 · similarity 0.86)
- Prompt-Driven Referring Image Segmentation with Instance Contrasting (CVPR 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.
