# Weakly-supervised continual learning for class-incremental segmentation

arXiv: https://arxiv.org/abs/2201.01029 · HTML: https://mlsanity.com/paper/2201.01029/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 10 of its 15 references that are in the map.
- First posted on arXiv 4 Jan 2022.

## Builds on (most cited first, 10 of 10)

- [LinkNet: Exploiting Encoder Representations for Efficient Semantic Segmentation](https://mlsanity.com/paper/1707.03718/) (2017)
- [Modeling the Background for Incremental Learning in Semantic Segmentation](https://mlsanity.com/paper/2002.00718/) (2020)
- [Continual Semantic Segmentation via Repulsion-Attraction of Sparse and Disentangled Latent Representations](https://mlsanity.com/paper/2103.06342/) (2021)
- [Incremental Learning for Semantic Segmentation of Large-Scale Remote Sensing Data](https://mlsanity.com/paper/1810.12448/) (2018)
- [Semantic Segmentation of Remote Sensing Images with Sparse Annotations](https://mlsanity.com/paper/2101.03492/) (2021)
- [Semi-Supervised Semantic Segmentation in Earth Observation: The MiniFrance Suite, Dataset Analysis and Multi-task Network Study](https://mlsanity.com/paper/2010.07830/) (2020)
- [Weakly Supervised Adversarial Domain Adaptation for Semantic Segmentation in Urban Scenes](https://mlsanity.com/paper/1904.09092/) (2019)
- [Weakly Supervised Change Detection Using Guided Anisotropic Difusion](https://mlsanity.com/paper/2112.15367/) (2021)
- [Interactive Learning for Semantic Segmentation in Earth Observation](https://mlsanity.com/paper/2009.11250/) (2020)
- [A Bayesian-inspired, deep learning-based, semi-supervised domain adaptation technique for land cover mapping](https://mlsanity.com/paper/2005.11930/) (2020)

## Built on by (most cited first, 2 of 2)

- [Artificial intelligence to advance Earth observation: : A review of models, recent trends, and pathways forward](https://mlsanity.com/paper/2305.08413/) (2023)
- [Continual Barlow Twins: continual self-supervised learning for remote sensing semantic segmentation](https://mlsanity.com/paper/2205.11319/) (2022)

## Abstract (arXiv)

Transfer learning is a powerful way to adapt existing deep learning models to new emerging use-cases in remote sensing. Starting from a neural network already trained for semantic segmentation, we propose to modify its label space to swiftly adapt it to new classes under weak supervision. To alleviate the background shift and the catastrophic forgetting problems inherent to this form of continual learning, we compare different regularization terms and leverage a pseudo-label strategy. We experimentally show the relevance of our approach on three public remote sensing datasets. Code is open-source and released in this repository: https://github.com/alteia-ai/ICSS}{https://github.com/alteia-ai/ICSS.