A deep multiple instance learning approach based on coarse labels for high-resolution land-cover mapping

2025-10-08

AI Summary

High-resolution land cover mapping faces a weakly supervised learning challenge due to the scarcity of pixel-level annotations, with only coarse-grained, low-resolution, or outdated labels (e.g., MODIS products) available. Method: This paper proposes a deep multiple-instance learning framework tailored for multi-class, multi-label scenarios, integrating high-resolution remote sensing imagery (e.g., Sentinel-2) with coarse reference labels. It introduces a learnable, flexible pooling mechanism and incorporates positive-unlabeled (PU) learning to implicitly model pixel-level semantics and mitigate class ambiguity. Contribution/Results: Experiments on the IEEE GRSS 2020 Data Fusion Contest dataset demonstrate that the method significantly outperforms conventional weakly supervised and fully supervised baselines—achieving high classification accuracy without any high-resolution ground truth annotations. This establishes a novel, cost-effective, and scalable paradigm for remote sensing semantic mapping.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Computer Vision: Remote Sensing / Geospatial AI
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

The quantity and the quality of the training labels are central problems in high-resolution land-cover mapping with machine-learning-based solutions. In this context, weak labels can be gathered in large quantities by leveraging on existing low-resolution or obsolete products. In this paper, we address the problem of training land-cover classifiers using high-resolution imagery (e.g., Sentinel-2) and weak low-resolution reference data (e.g., MODIS -derived land-cover maps). Inspired by recent works in Deep Multiple Instance Learning (DMIL), we propose a method that trains pixel-level multi-class classifiers and predicts low-resolution labels (i.e., patch-level classification), where the actual high-resolution labels are learned implicitly without direct supervision. This is achieved with flexible pooling layers that are able to link the semantics of the pixels in the high-resolution imagery to the low-resolution reference labels. Then, the Multiple Instance Learning (MIL) problem is re-framed in a multi-class and in a multi-label setting. In the former, the low-resolution annotation represents the majority of the pixels in the patch. In the latter, the annotation only provides us information on the presence of one of the land-cover classes in the patch and thus multiple labels can be considered valid for a patch at a time, whereas the low-resolution labels provide us only one label. Therefore, the classifier is trained with a Positive-Unlabeled Learning (PUL) strategy. Experimental results on the 2020 IEEE GRSS Data Fusion Contest dataset show the effectiveness of the proposed framework compared to standard training strategies.

Problem

• Training land-cover classifiers using high-resolution imagery with weak labels
• Linking high-resolution pixel semantics to low-resolution reference labels
• Addressing multi-class and multi-label classification with Positive-Unlabeled Learning

Innovation

• Deep Multiple Instance Learning for land-cover mapping
• Flexible pooling links high-resolution pixels to low-resolution labels
• Positive-Unlabeled Learning strategy handles multi-label classification