Research map / Detection and segmentation / CNN features
PanopticDepth: A Unified Framework for Depth-Aware Panoptic Segmentation
Check 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 |
|---|---|---|---|
| Use instance-wise dynamic-kernel depth estimation instead of pixel-wise depth regression (ablation) | PanopticDepth · Cityscapes-DPS | DPQ: 54.9 -> 54.9 | No clear effect |
| Use instance-wise dynamic-kernel depth estimation instead of pixel-wise depth regression (ablation) | PanopticDepth · SemKITTI-DPS | DPQ: 45.3 -> 45.8 | Helped |
| Add instance depth normalization using T1 (ablation) | PanopticDepth with instance-wise depth estimation · Cityscapes-DPS | DPQ: 54.9 -> 56.1 | Helped |
| Add instance depth normalization using T2 (ablation) | PanopticDepth with instance-wise depth estimation · SemKITTI-DPS | DPQ: 45.8 -> 46.7 | Helped |
| Add instance-level depth loss supervising depth shift with instance minimum depth (ablation) | PanopticDepth with T1 normalization · Cityscapes-DPS | DPQ: 56.1 -> 46.2 | Hurt |
| Add instance-level depth loss supervising depth shift with instance minimum depth (ablation) | PanopticDepth with T1 normalization · SemKITTI-DPS | DPQ: 46.7 -> 36.1 | Hurt |
| Add instance-level depth loss supervising depth shift with instance mean depth (ablation) | PanopticDepth with T2 normalization · Cityscapes-DPS | DPQ: 56.1 -> 57.0 | Helped |
| Add instance-level depth loss supervising depth shift with instance mean depth (ablation) | PanopticDepth with T2 normalization · SemKITTI-DPS | DPQ: 46.7 -> 46.9 | Helped |
| Use T2 normalization instead of T1 normalization (ablation) | PanopticDepth with instance-wise depth estimation, without instance-level depth loss · Cityscapes-DPS | DPQ: 56.1 -> 56.1 | No clear effect |
| Use instance mean depth rather than minimum depth as the target for instance-level depth shift supervision (ablation) | PanopticDepth with instance-level depth loss · Cityscapes-DPS | DPQ^Th: 47.2 -> 59.2 | Helped |
| Add adaptive kernel fusion (ablation) | PanopticFCN baseline implementation · Cityscapes val | PQ: 62.4 -> 63.6 | Helped |
| Fine-tune with full-scale images after adaptive kernel fusion (ablation) | PanopticFCN baseline implementation · Cityscapes val | PQ: 63.6 -> 64.1 | Helped |
Closest papers by meaning
- Real-Time Panoptic Segmentation From Dense Detections (CVPR 2020 · similarity 0.91)
- VIP-DeepLab: Learning Visual Perception With Depth-Aware Video Panoptic Segmentation (CVPR 2021 · similarity 0.91)
- SDC-Depth: Semantic Divide-and-Conquer Network for Monocular Depth Estimation (CVPR 2020 · similarity 0.90)
- Unifying Training and Inference for Panoptic Segmentation (CVPR 2020 · similarity 0.90)
- Scene-Centric Unsupervised Panoptic Segmentation (CVPR 2025 · similarity 0.88)
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.88)
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.