# Revisiting Continual Semantic Segmentation with Pre-trained Vision Models

arXiv: https://arxiv.org/abs/2508.04267 · HTML: https://mlsanity.com/paper/2508.04267/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 46 of its 59 references that are in the map.
- First posted on arXiv 6 Aug 2025.

## Builds on (most cited first, 30 of 46)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021)
- [Segment Anything](https://mlsanity.com/paper/2304.02643/) (2023)
- [Fully Convolutional Networks for Semantic Segmentation](https://mlsanity.com/paper/1411.4038/) (2014)
- [Overcoming catastrophic forgetting in neural networks](https://mlsanity.com/paper/1612.00796/) (2016)
- [DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs](https://mlsanity.com/paper/1606.00915/) (2016)
- [Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation](https://mlsanity.com/paper/1802.02611/) (2018)
- [Learning without Forgetting](https://mlsanity.com/paper/1606.09282/) (2016)
- [iCaRL: Incremental Classifier and Representation Learning](https://mlsanity.com/paper/1611.07725/) (2016)
- [A Comprehensive Survey of Continual Learning: Theory, Method and Application](https://mlsanity.com/paper/2302.00487/) (2023)
- [Large Scale Incremental Learning](https://mlsanity.com/paper/1905.13260/) (2019)
- [Recent Advances in Open Set Recognition: A Survey](https://mlsanity.com/paper/1811.08581/) (2018)
- [Modeling the Background for Incremental Learning in Semantic Segmentation](https://mlsanity.com/paper/2002.00718/) (2020)
- [PLOP: Learning without Forgetting for Continual Semantic Segmentation](https://mlsanity.com/paper/2011.11390/) (2020)
- [Incremental Learning Techniques for Semantic Segmentation](https://mlsanity.com/paper/1907.13372/) (2019)
- [Forward Compatible Few-Shot Class-Incremental Learning](https://mlsanity.com/paper/2203.06953/) (2022)
- [A Review on Deep Learning Techniques Applied to Semantic Segmentation](https://mlsanity.com/paper/1704.06857/) (2017)
- [No More Discrimination: Cross City Adaptation of Road Scene Segmenters](https://mlsanity.com/paper/1704.08509/) (2017)
- [SegViT: Semantic Segmentation with Plain Vision Transformers](https://mlsanity.com/paper/2210.05844/) (2022)
- [Continual Semantic Segmentation via Repulsion-Attraction of Sparse and Disentangled Latent Representations](https://mlsanity.com/paper/2103.06342/) (2021)
- [Contextual Transformer Networks for Visual Recognition](https://mlsanity.com/paper/2107.12292/) (2021)
- [RECALL: Replay-based Continual Learning in Semantic Segmentation](https://mlsanity.com/paper/2108.03673/) (2021)
- [Lifelong Learning of Large Language Model based Agents: A Roadmap](https://mlsanity.com/paper/2501.07278/) (2025)
- [Conv2Former: A Simple Transformer-Style ConvNet for Visual Recognition](https://mlsanity.com/paper/2211.11943/) (2022)
- [Probing Representation Forgetting in Supervised and Unsupervised Continual Learning](https://mlsanity.com/paper/2203.13381/) (2022)
- [Representation Compensation Networks for Continual Semantic Segmentation](https://mlsanity.com/paper/2203.05402/) (2022)
- [SSUL: Semantic Segmentation with Unknown Label for Exemplar-based Class-Incremental Learning](https://mlsanity.com/paper/2106.11562/) (2021)
- [Towards Lifelong Learning of Large Language Models: A Survey](https://mlsanity.com/paper/2406.06391/) (2024)

## Abstract (arXiv)

Continual Semantic Segmentation (CSS) seeks to incrementally learn to segment novel classes while preserving knowledge of previously encountered ones. Recent advancements in CSS have been largely driven by the adoption of Pre-trained Vision Models (PVMs) as backbones. Among existing strategies, Direct Fine-Tuning (DFT), which sequentially fine-tunes the model across classes, remains the most straightforward approach. Prior work often regards DFT as a performance lower bound due to its presumed vulnerability to severe catastrophic forgetting, leading to the development of numerous complex mitigation techniques. However, we contend that this prevailing assumption is flawed. In this paper, we systematically revisit forgetting in DFT across two standard benchmarks, Pascal VOC 2012 and ADE20K, under eight CSS settings using two representative PVM backbones: ResNet101 and Swin-B. Through a detailed probing analysis, our findings reveal that existing methods significantly underestimate the inherent anti-forgetting capabilities of PVMs. Even under DFT, PVMs retain previously learned knowledge with minimal forgetting. Further investigation of the feature space indicates that the observed forgetting primarily arises from the classifier's drift away from the PVM, rather than from degradation of the backbone representations. Based on this insight, we propose DFT*, a simple yet effective enhancement to DFT that incorporates strategies such as freezing the PVM backbone and previously learned classifiers, as well as pre-allocating future classifiers. Extensive experiments show that DFT* consistently achieves competitive or superior performance compared to sixteen state-of-the-art CSS methods, while requiring substantially fewer trainable parameters and less training time.