Research map / Video / Video segmentation and tracking
DIMOS: Disentangling Instance-level Moving Object Segmentation
Hongxiang Huang, Hongwei Ren, Xiaopeng Lin, Yulong Huang, Zeke Xie, Bojun Cheng
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 unsupervised optical flow loss (ablation) | DIMOS baseline · MouseSIS | mIoU_ins (%): 60.47 -> 62.54 | Improved |
| add bounding box regression loss (ablation) | DIMOS + flow · MouseSIS | mIoU_ins (%): 62.54 -> 63.46 | Improved |
| add dual-disentangling mechanism with intra-modal contrastive learning (ablation) | DIMOS + flow + bbox · MouseSIS | mIoU_ins (%): 63.46 -> 68.11 | Improved |
| add semantic-level alignment (modality translation) (ablation) | DIMOS + dual · MouseSIS | mIoU_ins (%): 68.11 -> 69.23 | Improved |
| add distribution-level alignment (adversarial) (ablation) | DIMOS + sem. align · MouseSIS | mIoU_ins (%): 69.23 -> 70.25 | Improved |
| use ResNet-50 instead of MobileNetV2 backbone (ablation) | DIMOS · MouseSIS | mIoU_ins (%): 68.62 -> 70.25 | Improved |
| use ResNet-50 instead of ResNet18 backbone (ablation) | DIMOS · MouseSIS | mIoU_ins (%): 69.32 -> 70.25 | Improved |
| stack 3 fusion blocks instead of 1 (ablation) | DIMOS · MouseSIS | mIoU_ins (%): 70.25 -> 70.23 | No clear effect |
| deepen domain discriminator to 3 blocks (ablation) | DIMOS · MouseSIS | mIoU_ins (%): 70.25 -> 69.52 | Hurt |
| use 3 modality translator blocks instead of 1 (ablation) | DIMOS · MouseSIS | mIoU_ins (%): 70.25 -> 70.11 | No clear effect |
| use DIMOS instead of EvInsMOS | DIMOS · SEVD-Fixed | mIoU_ins (%): 56.50 -> 62.05 | Improved |
| use DIMOS instead of EvInsMOS | DIMOS · MouseSIS | mAP (%): 30.94 -> 45.18 | Improved |
Closest papers by meaning
- Event-assisted Low-Light Video Object Segmentation (CVPR 2024 · similarity 0.88)
- Know Where You Are From: Event-Based Segmentation via Spatio-Temporal Propagation (AAAI 2025 · similarity 0.87)
- TrackIME: Enhanced Video Point Tracking via Instance Motion Estimation (NeurIPS 2024 · similarity 0.86)
- Segment Any Motion in Videos (CVPR 2025 · similarity 0.86)
- ESEG: Event-Based Segmentation Boosted by Explicit Edge-Semantic Guidance (AAAI 2025 · similarity 0.86)
- TimeTracker: Event-based Continuous Point Tracking for Video Frame Interpolation with Non-linear Motion (CVPR 2025 · 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.
