# Auto-Compressing Subset Pruning for Semantic Image Segmentation

arXiv: https://arxiv.org/abs/2201.11103 · HTML: https://mlsanity.com/paper/2201.11103/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 26 of its 42 references that are in the map.
- First posted on arXiv 26 Jan 2022.

## Builds on (most cited first, 26 of 26)

- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [The Cityscapes Dataset for Semantic Urban Scene Understanding](https://mlsanity.com/paper/1604.01685/) (2016)
- [Pyramid Scene Parsing Network](https://mlsanity.com/paper/1612.01105/) (2016)
- [The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks](https://mlsanity.com/paper/1803.03635/) (2018)
- [Pruning Filters for Efficient ConvNets](https://mlsanity.com/paper/1608.08710/) (2016)
- [Segmenter: Transformer for Semantic Segmentation](https://mlsanity.com/paper/2105.05633/) (2021)
- [Learning Efficient Convolutional Networks through Network Slimming](https://mlsanity.com/paper/1708.06519/) (2017)
- [What is the State of Neural Network Pruning?](https://mlsanity.com/paper/2003.03033/) (2020)
- [Predicting Parameters in Deep Learning](https://mlsanity.com/paper/1306.0543/) (2013)
- [A survey of loss functions for semantic segmentation](https://mlsanity.com/paper/2006.14822/) (2020)
- [A Constructive Prediction of the Generalization Error Across Scales](https://mlsanity.com/paper/1909.12673/) (2019)
- [Gate Decorator: Global Filter Pruning Method for Accelerating Deep Convolutional Neural Networks](https://mlsanity.com/paper/1909.08174/) (2019)
- [The Early Phase of Neural Network Training](https://mlsanity.com/paper/2002.10365/) (2020)
- [Data-free parameter pruning for Deep Neural Networks](https://mlsanity.com/paper/1507.06149/) (2015)
- [Stochastic Optimization of Sorting Networks via Continuous Relaxations](https://mlsanity.com/paper/1903.08850/) (2019)
- [Fast ConvNets Using Group-wise Brain Damage](https://mlsanity.com/paper/1506.02515/) (2015)
- [AutoPruner: An End-to-End Trainable Filter Pruning Method for Efficient Deep Model Inference](https://mlsanity.com/paper/1805.08941/) (2018)
- [An Overview of Neural Network Compression](https://mlsanity.com/paper/2006.03669/) (2020)
- [Differentiable Subset Pruning of Transformer Heads](https://mlsanity.com/paper/2108.04657/) (2021)
- [Batch-Shaping for Learning Conditional Channel Gated Networks](https://mlsanity.com/paper/1907.06627/) (2019)
- [Differentiable Top-k Operator with Optimal Transport](https://mlsanity.com/paper/2002.06504/) (2020)
- [Data Agnostic Filter Gating for Efficient Deep Networks](https://mlsanity.com/paper/2010.15041/) (2020)
- [Plug-in, Trainable Gate for Streamlining Arbitrary Neural Networks](https://mlsanity.com/paper/1904.10921/) (2019)
- [LEAN: graph-based pruning for convolutional neural networks by extracting longest chains](https://mlsanity.com/paper/2011.06923/) (2020)
- [GPU Acceleration of Sparse Neural Networks](https://mlsanity.com/paper/2005.04347/) (2020)

## Built on by (most cited first, 4 of 4)

- [Differentiable Transportation Pruning](https://mlsanity.com/paper/2307.08483/) (2023)
- [Sauron U-Net: Simple automated redundancy elimination in medical image segmentation via filter pruning](https://mlsanity.com/paper/2209.13590/) (2022)
- [Optimization of DNN-based HSI Segmentation FPGA-based SoC for ADS: A Practical Approach](https://mlsanity.com/paper/2507.16556/) (2025)
- [StaticSegFormer: An Efficient High-Performance Semantic Segmentation Based on Static Structured Pruning](https://mlsanity.com/paper/2608.04811/) (2026)

## Abstract (arXiv)

State-of-the-art semantic segmentation models are characterized by high parameter counts and slow inference times, making them unsuitable for deployment in resource-constrained environments. To address this challenge, we propose textscAuto-Compressing Subset Pruning, acosp, as a new online compression method. The core of acosp consists of learning a channel selection mechanism for individual channels of each convolution in the segmentation model based on an effective temperature annealing schedule. We show a crucial interplay between providing a high-capacity model at the beginning of training and the compression pressure forcing the model to compress concepts into retained channels. We apply acosp to segnet and pspnet architectures and show its success when trained on the camvid, city, voc, and ade datasets. The results are competitive with existing baselines for compression of segmentation models at low compression ratios and outperform them significantly at high compression ratios, yielding acceptable results even when removing more than 93% of the parameters. In addition, acosp is conceptually simple, easy to implement, and can readily be generalized to other data modalities, tasks, and architectures. Our code is available at urlhttps://github.com/merantix/acosp.