Research map / Label-efficient and robust learning / Domain adaptation and generalization
Learning Across Tasks and Domains
Pierluigi Zama Ramirez, Alessio Tonioni, Samuele Salti, Luigi Di Stefano
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 a single task-1 network shared across both domains (ablation) | AT/DT · Synthia->Cityscapes, depth-to-semantics (target) | mIoU: 6.42 -> 23.24 (+16.82) | Improved |
| add a single shared task-1 network (ablation) | AT/DT · Synthia->Cityscapes (target) | accuracy: 29.36 -> 64.03 (+34.67) | Improved |
| add a single shared task-1 network (ablation) | AT/DT · Synthia (source domain A) | mIoU: 61.73 -> 65.41 | Improved |
| add batch normalization to the task networks (ablation) | AT/DT · Synthia->Cityscapes, target domain | accuracy: 58.29 -> 64.03 (+5.74) | Improved |
| add batch normalization to the task networks (ablation) | AT/DT · Synthia->Cityscapes, target domain | mIoU: 22.75 -> 23.24 (+0.49) | Improved |
| add batch normalization to the task networks (ablation) | AT/DT · Synthia (source domain A) | mIoU: 72.48 -> 65.41 | Hurt |
| transfer features at deeper layers of the network (ablation) | AT/DT · Synthia->Cityscapes, depth-to-semantics | Acc and mIoU | Improved |
| add AT/DT cross-task transfer over the baseline | task-2 network trained on A · Synthia->Carla, depth-to-semantics | mIoU: 34.12 -> 43.14 | Improved |
| add AT/DT cross-task transfer over the baseline | task-2 network trained on A · Synthia->Cityscapes, depth-to-semantics | mIoU: 7.72 -> 23.24 | Improved |
| add AT/DT cross-task transfer over the baseline | task-2 network trained on A · Carla->Cityscapes, semantics-to-depth | Abs Rel: 0.667 -> 0.394 | Improved |
| add AT/DT cross-task transfer over the baseline | task-2 network trained on A · Carla->Kitti, semantics-to-depth | Abs Rel: 0.500 -> 0.439 | Improved |
| add CycleGAN pixel-level domain adaptation | baseline · Carla->Cityscapes, depth-to-semantics | mIoU: 37.88 -> 42.33 | Improved |
Closest papers by meaning
- Learning Good Features to Transfer Across Tasks and Domains (TPAMI 2023 · similarity 0.92)
- Learning To Relate Depth and Semantics for Unsupervised Domain Adaptation (CVPR 2021 · similarity 0.90)
- Self-Supervised Learning Across Domains (TPAMI 2021 · similarity 0.89)
- Taskonomy: Disentangling Task Transfer Learning (CVPR 2018 · similarity 0.89)
- Factors of Influence for Transfer Learning Across Diverse Appearance Domains and Task Types (TPAMI 2021 · similarity 0.89)
- A General Representation-Based Approach to Multi-Source Domain Adaptation (ICML 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.
