Research map / Detection and segmentation / CNN features
LPSNet: A Lightweight Solution for Fast 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 |
|---|---|---|---|
| Increase the train/test image shorter side from 800 to 964 | LPSNet · COCO val | PQ: 39.1 -> 42.4 | Helped |
| Use LPSNet rather than UPSNet at image size 800 | COCO val · COCO val | time: 202 -> 108 | Helped |
| Remove Pyramid Pooling Module (ablation) | LPSNet · COCO val | mIoU: 54.5 -> 50.9 | Hurt |
| Add Pyramid Pooling Module (ablation) | LPSNet · COCO val | PQ: 35.7 -> 39.1 | Helped |
| Replace PPM with Global Average Pooling (ablation) | LPSNet · COCO val | mIoU: 54.5 -> 51.2 | Hurt |
| Use closest Border-ness assignment instead of smallest-area assignment (ablation) | LPSNet panoptic head · COCO val | PQ: 38.9 -> 39.1 | Helped |
| Use closest Border-ness assignment instead of highest-confidence assignment (ablation) | LPSNet panoptic head · COCO val | PQ: 39.0 -> 39.1 | Helped |
| Replace predicted semantic segmentation with ground-truth labels (ablation) | UPSNet · COCO val | PQ: 42.5 -> 72.0 | Helped |
| Replace predicted semantic segmentation with ground-truth labels (ablation) | LPSNet · COCO val | PQ: 39.1 -> 74.1 | Helped |
| Replace predicted boxes with ground-truth boxes (ablation) | UPSNet · COCO val | PQ: 42.5 -> 60.8 | Helped |
| Replace predicted boxes with ground-truth boxes (ablation) | LPSNet · COCO val | PQ: 39.1 -> 53.0 | Helped |
| Use mask-based positive samples rather than the FCOS box-based definition (ablation) | LPSNet box head · COCO | PQ | Helped |
Closest papers by meaning
- Real-Time Panoptic Segmentation From Dense Detections (CVPR 2020 · similarity 0.91)
- Unifying Training and Inference for Panoptic Segmentation (CVPR 2020 · similarity 0.91)
- Fully Convolutional Networks for Panoptic Segmentation (CVPR 2021 · similarity 0.91)
- Combinatorial Optimization for Panoptic Segmentation: A Fully Differentiable Approach (NeurIPS 2021 · similarity 0.90)
- UPSNet: A Unified Panoptic Segmentation Network (CVPR 2019 · similarity 0.90)
- Scene-Centric Unsupervised Panoptic Segmentation (CVPR 2025 · similarity 0.87)
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.