mlsanityData to 7 Oct 2026

Class Based Thresholding in Early Exit Semantic Segmentation Networks

arXiv 2210.15621 ↗First posted 27 Oct 2022cs.CV cs.LGcited by 5

On the map: Efficient LLM inference and serving → Token pruning and merging for ViTs

Abstract (arXiv)

We propose Class Based Thresholding (CBT) to reduce the computational cost of early exit semantic segmentation models while preserving the mean intersection over union (mIoU) performance. A key idea of CBT is to exploit the naturally-occurring neural collapse phenomenon. Specifically, by calculating the mean prediction probabilities of each class in the training set, CBT assigns different masking threshold values to each class, so that the computation can be terminated sooner for pixels belonging to easy-to-predict classes. We show the effectiveness of CBT on Cityscapes and ADE20K datasets. CBT can reduce the computational cost by 23% compared to the previous state-of-the-art early exit models.

Key facts. Cited by 5 papers in the map (data through 7 Oct 2026). Placed in Token pruning and merging for ViTs, one of 12 directions in Efficient LLM inference and serving. Builds on 23 of its 47 references that are in the map. First posted on arXiv 27 Oct 2022.

Builds on

23 of 47 references.

References · 23

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  3. 2015Distilling the Knowledge in a Neural Networkcited by 10,076
  4. 2016The Cityscapes Dataset for Semantic Urban Scene Understandingcited by 4,427
  5. 2022OPT: Open Pre-trained Transformer Language Modelscited by 3,464
  6. 2021Masked-attention Mask Transformer for Universal Image Segmentationcited by 2,338
  7. 2015Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Codingcited by 2,291
  8. 2019Deep High-Resolution Representation Learning for Visual Recognitioncited by 1,417
  9. 2016COCO-Stuff: Thing and Stuff Classes in Contextcited by 858
  10. 2020Prevalence of Neural Collapse during the terminal phase of deep learning trainingcited by 713
  11. 2017BranchyNet: Fast Inference via Early Exiting from Deep Neural Networkscited by 591
  12. 2022Compute Trends Across Three Eras of Machine Learningcited by 161
11 more
  1. 2020Multiscale Deep Equilibrium Modelscited by 149
  2. 2021Standardized Max Logits: A Simple yet Effective Approach for Identifying Unexpected Road Obstacles in Urban-Scene Segmentationcited by 77
  3. 2022Limitations of Neural Collapse for Understanding Generalization in Deep Learningcited by 63
  4. 2021CondenseNet V2: Sparse Feature Reactivation for Deep Networkscited by 34
  5. 2018Learn to Combine Modalities in Multimodal Deep Learningcited by 29
  6. 2021Anytime Dense Prediction with Confidence Adaptivitycited by 18
  7. 2021Multi-Exit Semantic Segmentation Networkscited by 18
  8. 2019Efficient Ladder-style DenseNets for Semantic Segmentation of Large Imagescited by 17
  9. 2021Dynamically pruning segformer for efficient semantic segmentationcited by 13
  10. 2021E2CM: Early Exit via Class Means for Efficient Supervised and Unsupervised Learningcited by 8
  11. 2022Pruning Early Exit Networkscited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 2; 2024: 1; 2025: 1; 2026: 1.

5 papers.

  1. 2025A Survey on Dynamic Neural Networks: from Computer Vision to Multi-modal Sensor Fusioncited by 5
  2. 2023Memorization Capacity of Neural Networks with Conditional Computationcited by 3
  3. 2024On the Impact of Black-box Deployment Strategies for Edge AI on Latency and Model Performancecited by 3
  4. 2023Gated Compression Layers for Efficient Always-On Modelscited by 2
  5. 2026Theoretical Guarantees for One-Shot Magnitude Pruning and Compute-Adaptive Early Exitcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2023Dynamic Token Pruning in Plain Vision Transformers for Semantic Segmentation
  2. 2024MaskDiffusion: Exploiting Pre-trained Diffusion Models for Semantic Segmentation
  3. 2024ALGM: Adaptive Local-then-Global Token Merging for Efficient Semantic Segmentation with Plain Vision Transformers
  4. 2025Training-Free Class Purification for Open-Vocabulary Semantic Segmentation
  5. 2024SOOD-ImageNet: a Large-Scale Dataset for Semantic Out-Of-Distribution Image Classification and Semantic Segmentation
  6. 2025Diffusion-Driven Two-Stage Active Learning for Low-Budget Semantic Segmentation
  7. 2022Decomposed Knowledge Distillation for Class-Incremental Semantic Segmentation
  8. 2023Early-Exit with Class Exclusion for Efficient Inference of Neural Networks