Research map / Label-efficient and robust learning / Domain adaptation and generalization
Occlusion-Aware Seamless Segmentation
Yihong Cao, Jiaming Zhang, Hao Shi, Kunyu Peng, Yuhongxuan Zhang, Hui Zhang, Rainer Stiefelhagen, Kailun Yang
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 Unmasking Attention (UA) backbone to the baseline (ablation) | UnmaskFormer · KITTI360-APS->BlendPASS | mAPQ: 22.00 -> 25.04 | Improved |
| add AoMix masked mixing (ablation) | UnmaskFormer · KITTI360-APS->BlendPASS | mAPQ: 22.00 -> 25.17 | Improved |
| combine UA and AoMix (ablation) | UnmaskFormer · KITTI360-APS->BlendPASS | mAPQ: 25.04 / 25.17 -> 26.55 | Improved |
| add OAFusion for fusing branch outputs (ablation) | UnmaskFormer · KITTI360-APS->BlendPASS | mAPQ: 26.55 -> 26.58 | Improved |
| interleave DPE in Stages 2 and 4 instead of early-stage DPE or plain PE (ablation) | UnmaskFormer · KITTI360-APS->BlendPASS | mAPQ: +3.04 and +2.14 gain | Improved |
| use GAP-based occlusion-aware pooling attention instead of SimPool (ablation) | UA · KITTI360-APS->BlendPASS | mAPQ: 24.74 -> 25.04 | Improved |
| mask Thing regions only instead of whole image or patches (ablation) | AoMix · KITTI360-APS->BlendPASS | mAPQ: +1.05 and +3.3 mAPQ | Improved |
| add more DPE blocks across stages (ablation) | UnmaskFormer · KITTI360-APS->BlendPASS | mAPQ / mIoU | No clear effect |
| use amodal instance output vs visible for EDAPS-comparable instance seg | UnmaskFormer · KITTI360-APS->BlendPASS | mAAP: 10.50 vs 10.68 (EDAPS) | No clear effect |
| overly large or small AoMix scaling range (ablation) | AoMix · KITTI360-APS->BlendPASS | mAPQ | Hurt |
Closest papers by meaning
- Segment Anything, Even Occluded (CVPR 2025 · similarity 0.86)
- PanoOcc: Unified Occupancy Representation for Camera-based 3D Panoptic Segmentation (CVPR 2024 · similarity 0.86)
- Amodal Scene Analysis via Holistic Occlusion Relation Inference and Generative Mask Completion (AAAI 2024 · similarity 0.86)
- Geometric Exploitation for Indoor Panoramic Semantic Segmentation (NeurIPS 2024 · similarity 0.85)
- Using Diffusion Priors for Video Amodal Segmentation (CVPR 2025 · similarity 0.85)
- Semantics Distortion and Style Matter: Towards Source-free UDA for Panoramic Segmentation (CVPR 2024 · 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.
