Research map / Detection and segmentation / CNN features
Fully Convolutional Networks for 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 generated kernels to represent things and stuff in one fully convolutional pipeline | Panoptic FCN · COCO val | PQ: 43.6 vs 43.4 | Helped |
| Add deformable convolutions to the position head (ablation) | Panoptic FCN kernel generator · COCO val | PQ: 39.2 -> 39.9 | Helped |
| Add coordinates to both kernel head and feature encoder (ablation) | Panoptic FCN · COCO val | PQ: 39.9 -> 41.3 | Helped |
| Set kernel-fusion cosine threshold to 1.00 instead of 0.90 (ablation) | Panoptic FCN · COCO val | PQ: 41.3 -> 38.7 | Hurt |
| Use class-agnostic rather than class-aware kernel fusion (ablation) | Panoptic FCN · COCO val | AP: 32.1 -> 30.9 | Hurt |
| Add kernel-level fusion instead of relying on NMS (ablation) | Panoptic FCN · COCO val | PQ: 38.7 -> 41.3 | Helped |
| Add NMS on top of kernel fusion (ablation) | Panoptic FCN · COCO val | PQ: 41.3 -> 41.3 | No clear effect |
| Use Semantic FPN features rather than FPN-P2 or summed FPN features (ablation) | Panoptic FCN feature encoder · COCO val | PQ: 40.6 -> 41.3; 40.5 -> 41.3 | Helped |
| Sample multiple high-scoring kernels and optimize with weighted Dice loss (ablation) | Panoptic FCN · COCO val | PQ: 40.2 -> 41.3 | Helped |
| Extend training from 1× to 2× schedule (ablation) | Panoptic FCN · COCO val | PQ: 41.3 -> 43.2 | Helped |
| Extend training from 2× to 3× schedule (ablation) | Panoptic FCN · COCO val | PQ: 43.2 -> 43.6 | Helped |
| Add the paper's simple enhancements to the default model (ablation) | Panoptic FCN · COCO val | PQ: 43.6 -> 44.3 | Helped |
Closest papers by meaning
- Unifying Training and Inference for Panoptic Segmentation (CVPR 2020 · similarity 0.92)
- LPSNet: A Lightweight Solution for Fast Panoptic Segmentation (CVPR 2021 · similarity 0.91)
- Real-Time Panoptic Segmentation From Dense Detections (CVPR 2020 · similarity 0.90)
- Attention-Guided Unified Network for Panoptic Segmentation (CVPR 2019 · similarity 0.90)
- Scene-Centric Unsupervised Panoptic Segmentation (CVPR 2025 · similarity 0.90)
- Panoptic Feature Pyramid Networks (CVPR 2019 · similarity 0.90)
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.