Research map / Detection and segmentation / CNN features
Distributed Semantic Segmentation with Efficient Joint Source and Task Decoding
Danish Nazir, Timo Bartels, Jan Piewek, Thorsten Bagdonat, Tim Fingscheidt
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 |
|---|---|---|---|
| replace separate source decoder and task decoder with a single joint source and task decoder | distributed DeepLabv3 ResNet-50 cloud DNN · COCO-2017 val | #params (M): 16.78 -> 1.66 | Improved |
| replace separate source decoder and task decoder with joint decoder | distributed DeepLabv3 ResNet-50 cloud DNN · Cityscapes val | #params (M): 42.44 -> 4.92 | Improved |
| joint decoding versus Ahuja et al. on rate-distortion mIoU, COCO | distributed semantic segmentation · COCO-2017 val | mIoU | Improved |
| joint decoding versus Ahuja et al. on rate-distortion mIoU, Cityscapes | distributed semantic segmentation · Cityscapes val | mIoU | No clear effect |
| add training-time over-parameterization in the ASPP block of the joint decoder (ablation) | joint decoder JD · COCO-2017 val and Cityscapes val | mIoU at high bitrates | Improved |
| increase ASPP feature maps F from 256 to 512 (ablation) | joint decoder JD · COCO-2017 val | mIoU | Mixed |
| use F=512 instead of 128 or 256 (ablation) | joint decoder JD · Cityscapes val | mIoU | Improved |
| increase over-parameterization repetitions K from 1 to 2 (ablation) | joint decoder JD · COCO-2017 val | mIoU | No clear effect |
| use K=3 over-parameterization repetitions (ablation) | joint decoder JD · Cityscapes val | mIoU | Improved |
Closest papers by meaning
- Temporally Distributed Networks for Fast Video Semantic Segmentation (CVPR 2020 · similarity 0.86)
- Deep FusionNet for Point Cloud Semantic Segmentation (ECCV 2020 · similarity 0.86)
- Lightweight and Progressively-Scalable Networks for Semantic Segmentation (IJCV 2023 · similarity 0.86)
- Semantic Bottlenecks: Quantifying and Improving Inspectability of Deep Representations (IJCV 2021 · similarity 0.86)
- Domain Adaptive and Generalizable Network Architectures and Training Strategies for Semantic Image Segmentation (TPAMI 2023 · similarity 0.86)
- Collaborating Foundation Models for Domain Generalized Semantic Segmentation (CVPR 2024 · similarity 0.84)
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.
