# When Noise Meets Long-Tail: Feature-Threshold Dual Calibration for Robust Pseudo-Labeling

arXiv: https://arxiv.org/abs/2609.33668 · HTML: https://mlsanity.com/paper/2609.33668/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 23 of its 29 references that are in the map.
- First posted on arXiv 27 Sep 2026.

## Builds on (most cited first, 23 of 23)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [The Cityscapes Dataset for Semantic Urban Scene Understanding](https://mlsanity.com/paper/1604.01685/) (2016)
- [FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence](https://mlsanity.com/paper/2001.07685/) (2020)
- [FlexMatch: Boosting Semi-Supervised Learning with Curriculum Pseudo Labeling](https://mlsanity.com/paper/2110.08263/) (2021)
- [Semi-Supervised Semantic Segmentation with Cross Pseudo Supervision](https://mlsanity.com/paper/2106.01226/) (2021)
- [FreeMatch: Self-adaptive Thresholding for Semi-supervised Learning](https://mlsanity.com/paper/2205.07246/) (2022)
- [Revisiting Weak-to-Strong Consistency in Semi-Supervised Semantic Segmentation](https://mlsanity.com/paper/2208.09910/) (2022)
- [CReST: A Class-Rebalancing Self-Training Framework for Imbalanced Semi-Supervised Learning](https://mlsanity.com/paper/2102.09559/) (2021)
- [Semi-Supervised Semantic Segmentation via Adaptive Equalization Learning](https://mlsanity.com/paper/2110.05474/) (2021)
- [Semantic Segmentation of Underwater Imagery: Dataset and Benchmark](https://mlsanity.com/paper/2004.01241/) (2020)
- [UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation](https://mlsanity.com/paper/2410.10777/) (2024)
- [DASO: Distribution-Aware Semantics-Oriented Pseudo-label for Imbalanced Semi-Supervised Learning](https://mlsanity.com/paper/2106.05682/) (2021)
- [SemiVL: Semi-Supervised Semantic Segmentation with Vision-Language Guidance](https://mlsanity.com/paper/2311.16241/) (2023)
- [Switching Temporary Teachers for Semi-Supervised Semantic Segmentation](https://mlsanity.com/paper/2310.18640/) (2023)
- [The Marine Debris Dataset for Forward-Looking Sonar Semantic Segmentation](https://mlsanity.com/paper/2108.06800/) (2021)
- [Beyond Pixels: Semi-Supervised Semantic Segmentation with a Multi-scale Patch-based Multi-Label Classifier](https://mlsanity.com/paper/2407.04036/) (2024)
- [Virtual Category Learning: A Semi-Supervised Learning Method for Dense Prediction with Extremely Limited Labels](https://mlsanity.com/paper/2312.01169/) (2023)
- [Pseudo-Label Noise Suppression Techniques for Semi-Supervised Semantic Segmentation](https://mlsanity.com/paper/2210.10426/) (2022)
- [Zero-Shot Learning in Industrial Scenarios: New Large-Scale Benchmark, Challenges and Baseline](https://mlsanity.com/paper/2606.07965/) (2026)
- [RSOD: Reliability-Guided Sonar Image Object Detection with Extremely Limited Labels](https://mlsanity.com/paper/2601.12715/) (2026)
- [CTFS : Collaborative Teacher Framework for Forward-Looking Sonar Image Semantic Segmentation with Extremely Limited Labels](https://mlsanity.com/paper/2603.21071/) (2026)
- [Semi-supervised Source Detection in Astronomical Images: New Benchmark and Strong Baseline](https://mlsanity.com/paper/2606.09219/) (2026)

## Built on by (most cited first, 1 of 1)

- [Benchmarking Answer-Grounding Consistency in Remote Sensing Visual Question Answering](https://mlsanity.com/paper/2610.13636/) (2026)

## Abstract (arXiv)

Pseudo-labeling has become a cornerstone of learning from unlabeled data in semantic segmentation. Yet its effectiveness drops sharply in real-world scenarios where strong imaging noise and long-tailed class distributions occur together. We trace this failure to a vicious cycle of pseudo-label degradation. Imaging noise entangles foreground and background features, lowering prediction confidence across all classes, while long-tailed distributions leave tail classes with far fewer training samples and inherently lower confidence. Under fixed high-threshold filtering, these tail-class predictions are systematically filtered out, so they receive no supervision from unlabeled data and thus features keep degrading in subsequent iterations. Critically, noise and long-tail are not independent obstacles but mutually amplifying ones, and addressing either alone is insufficient. To break this cycle, we propose FTC-Seg, a Feature-Threshold dual-Calibration framework built on a standard teacher-student framework. At the feature level, Orthogonal Prototype Reconstruction (OPR) uses a set of learnable orthogonal prototypes to residually purify pixel-wise features, widening the margin between weak foreground targets and noisy backgrounds. At the threshold level, Adaptive Threshold Calibration (ATC) dynamically adjusts class-specific thresholds based on learning difficulty and prediction-distribution bias, rescuing low-confidence pseudo-labels of tail classes from systematic exclusion. Extensive experiments on four public benchmarks spanning three distinct noise modalities show that FTC-Seg achieves strong performance against state-of-the-art methods, with particularly substantial gains on tail classes. Our results establish that jointly calibrating features and thresholds is essential for robust pseudo-labeling under compounded noise and class imbalance.