Research map / Label-efficient and robust learning / Semi-supervised learning
Naive-Student: Leveraging Semi-Supervised Learning in Video Sequences for Urban Scene Segmentation
Liang-Chieh Chen, Raphael Gontijo Lopes, Bowen Cheng, Maxwell D. Collins, Ekin D. Cubuk, Barret Zoph, Hartwig Adam, Jonathon Shlens
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 pseudo-labeled video frames (train-sequence + train-extra) and WR-41 student | Panoptic-DeepLab X-71 · Cityscapes val-fine | PQ (%): 65.3 -> 69.1 | Improved |
| iterate once more with WR-41 teacher | Naive-Student · Cityscapes val-fine | PQ (%): 69.1 -> 70.4 | Improved |
| initialize student from iteration 1 checkpoint (ablation) | Naive-Student iteration 2 · Cityscapes val-fine | PQ (%): 70.4 -> 70.8 | Improved |
| train longer (60K to 120K iterations) (ablation) | Panoptic-DeepLab on train-fine · Cityscapes | PQ (%): 62.9 -> 62.7 | No clear effect |
| add pseudo-labeled train-sequence instead of train-fine (ablation) | Student X-71 · Cityscapes val | PQ (%): 63.1 -> 65.3 | Improved |
| mix human-labeled train-fine with pseudo-labeled train-sequence (ablation) | Student X-71 · Cityscapes val | PQ (%): 65.3 -> 65.2 | Hurt |
| add pseudo-labeled train-extra (ablation) | Student X-71 · Cityscapes val | PQ (%): 65.3 -> 66.9 | Improved |
| add Void-Ego-Car assignment in teacher (ablation) | Teacher pseudo-labels · Cityscapes val | PQ (%): 66.8 -> 67.5 | Improved |
| add test-time augmentation to teacher (ablation) | Teacher pseudo-labels · Cityscapes val | PQ (%): 66.7 -> 67.5 | Improved |
| add val-fine human labels to teacher training (ablation) | Teacher pseudo-labels · Cityscapes val | PQ (%): 66.1 -> 67.5 | Improved |
| add Mapillary Vistas pretraining to teacher (ablation) | Teacher pseudo-labels · Cityscapes val | PQ (%): 65.3 -> 67.5 | Improved |
| add semi-supervised pseudo-label training (ablation) | WR-41 · Cityscapes val | PQ (%): 64.0 -> 69.1 | Improved |
Closest papers by meaning
- UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation (TPAMI 2025 · similarity 0.90)
- Semi-Supervised Semantic Segmentation With High- and Low-Level Consistency (TPAMI 2019 · similarity 0.90)
- Semi-Supervised Segmentation based on Error-Correcting Supervision (ECCV 2020 · similarity 0.90)
- Improving Semi-Supervised and Domain-Adaptive Semantic Segmentation with Self-Supervised Depth Estimation (IJCV 2023 · similarity 0.89)
- Big Self-Supervised Models are Strong Semi-Supervised Learners (NeurIPS 2020 · similarity 0.89)
- AllSpark: Reborn Labeled Features from Unlabeled in Transformer for Semi-Supervised Semantic Segmentation (CVPR 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.
