Research map / Detection and segmentation / Open-set and few-shot detection
Reinforcement Cutting-Agent Learning for Video Object Segmentation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 a box-location channel to the CEN input (ablation) | CEN · DAVIS | Jm: 67.5 -> 73.9; 71.0 -> 76.8 | Helped |
| Add frame cropping using the preceding frame's object box and context (ablation) | CEN · DAVIS | Jm: 67.5 -> 71.0; 73.9 -> 76.8 | Helped |
| Learn object and context boxes jointly rather than using a previous-frame box and fixed context ratio (ablation) | reinforcement cutting-agent · DAVIS | Jm: 76.8 -> 83.9 | Helped |
| Add object searching actions that refine the object box (ablation) | reinforcement cutting-agent · DAVIS | Jm: 78.6 -> 83.9 | Helped |
| Add context embedding actions that select the context box (ablation) | reinforcement cutting-agent · DAVIS | Jm: 80.7 -> 83.9 | Helped |
| Fine-tune on the annotated first video frame (ablation) | OURS · DAVIS | Jm: 78.7 -> 83.9 | Helped |
| Use FC-DenseNet67 instead of FC-DenseNet56 as CEN (ablation) | OURS · DAVIS | Jm: 83.9 -> 85.4 | Helped |
| Use the proposed cutting-agent framework rather than MSK | video object segmentation · DAVIS | Jm: 80.3 -> 83.9 | Helped |
| Use the proposed cutting-agent framework rather than OnAVOS | video object segmentation · DAVIS validation | Jm: 86.1 -> 84.1 | Hurt |
| Use the proposed cutting-agent framework rather than OFL | video object segmentation · YouTube-Objects | mean J: 77.6 -> 78.1 | Helped |
Closest papers by meaning
- Actor-Critic Instance Segmentation (CVPR 2019 · similarity 0.88)
- Fast Template Matching and Update for Video Object Tracking and Segmentation (CVPR 2020 · similarity 0.88)
- Learning To Recommend Frame for Interactive Video Object Segmentation in the Wild (CVPR 2021 · similarity 0.88)
- Focus-Then-Decide: Segmentation-Assisted Reinforcement Learning (AAAI 2024 · similarity 0.86)
- LENS: Learning to Segment Anything with Unified Reinforced Reasoning (AAAI 2026 · similarity 0.86)
- SAM-R1: Leveraging SAM for Reward Feedback in Multimodal Segmentation via Reinforcement Learning (NeurIPS 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.