Cloud formations often obscure optical satellite-based monitoring of the Earth’s surface, thus limiting Earth observation (EO) activities such as land cover mapping, ocean color analysis, and cropland monitoring. The integration of machine learning (ML) methods within the remote sensing domain has significantly improved performance for a wide range of EO tasks, including cloud detection and filtering, but there is still much room for improvement. A key bottleneck is that ML methods typically depend on large amounts of annotated data for training, which are often difficult to come by in EO contexts. This is especially true when it comes to cloud optical thickness (COT) estimation. A reliable estimation of COT enables more fine-grained and application-dependent control compared to using pre-specified cloud categories, as is common practice. To alleviate the COT data scarcity problem, in this work, we propose a novel synthetic dataset for COT estimation, which we subsequently leverage for obtaining reliable and versatile cloud masks on real data. In our dataset, top-of-atmosphere radiances have been simulated for 12 of the spectral bands of the Multispectral Imagery (MSI) sensor onboard Sentinel-2 platforms. These data points have been simulated under consideration of different cloud types, COTs, and ground surface and atmospheric profiles. Extensive experimentation of training several ML models to predict COT from the measured reflectivity of the spectral bands demonstrates the usefulness of our proposed dataset. In particular, by thresholding COT estimates from our ML models, we show on two satellite image datasets (one that is publicly available, and one which we have collected and annotated) that reliable cloud masks can be obtained. The synthetic data, the newly collected real dataset, code and models have been made publicly available.

This paper presents a Convolutional Neural Networks (CNN) based Unsupervised Curriculum Learning approach for the recognition of water bodies to overcome the stated challenges for remote sensing based RGB imagery. The unsupervised nature of the presented algorithm eliminates the need for labelled training data. The problem is cast as a two class clustering problem (water and non-water), while clustering is done on deep features obtained by a pre-trained CNN. After initial clusters have been identified, representative samples from each cluster are chosen by the unsupervised curriculum learning algorithm for fine-tuning the feature extractor. The stated process is repeated iteratively until convergence. Three datasets have been used to evaluate the approach and show its effectiveness on varying scales: (i) SAT-6 dataset comprising high resolution aircraft images, (ii) Sentinel-2 of EuroSAT, comprising remote sensing images with low resolution, and (iii) PakSAT, a new dataset we created for this study. PakSAT is the first Pakistani Sentinel-2 dataset designed to classify water bodies of Pakistan. Extensive experiments on these datasets demonstrate the progressive learning behaviour of UCL and reported promising results of water classification on all three datasets. The obtained accuracies outperform the supervised methods in domain adaptation, demonstrating the effectiveness of the proposed algorithm.

Seagrass ecosystems are pivotal in marine environments, serving as crucial habitats for diverse marine species and contributing significantly to carbon sequestration. Accurate classification of seagrass species from underwater images is imperative for monitoring and preserving these ecosystems. This paper introduces Unsupervised Curriculum Learning (UCL) to seagrass classification using the DeepSeagrass dataset. UCL progressively learns from simpler to more complex examples, enhancing the model's ability to discern seagrass features in a curriculum-driven manner. Experiments employing state-of-the-art deep learning architectures, convolutional neural networks (CNNs), show that UCL achieved overall 90.12 % precision and 89 % recall, which significantly improves classification accuracy and robustness, outperforming some traditional supervised learning approaches like SimCLR, and unsupervised approaches like Zero-shot CLIP. The methodology of UCL involves four main steps: high-dimensional feature extraction, pseudo-label generation through clustering, reliable sample selection, and fine-tuning the model. The iterative UCL framework refines CNN's learning of underwater images, demonstrating superior accuracy, generalization, and adaptability to unseen seagrass and background samples of undersea images. The findings presented in this paper contribute to the advancement of seagrass classification techniques, providing valuable insights into the conservation and management of marine ecosystems. The code and dataset are made publicly available and can be assessed here: https://github.com/nabid69/Unsupervised-Curriculum-Learning—UCL.

Massive wildfires not only in Australia, but also worldwide are burning millions of hectares of forests and green land affecting the social, ecological, and economical situation. Widely used indices-based threshold methods like Normalized Burned Ratio (NBR) require a huge amount of data preprocessing and are specific to the data capturing source. State-of-the-art deep learning models, on the other hand, are supervised and require domain experts knowledge for labeling the data in huge quantity. These limitations make the existing models difficult to be adaptable to new variations in the data and capturing sources. In this work, we have proposed an unsupervised deep learning based architecture to map the burnt regions of forests by learning features progressively. The model considers small patches of satellite imagery and classifies them into burnt and not burnt. These small patches are concatenated into binary masks to segment out the burnt region of the forests. The proposed system is composed of two modules: 1) a state-of-the-art deep learning architecture for feature extraction and 2) a clustering algorithm for the generation of pseudo labels to train the deep learning architecture. The proposed method is capable of learning the features progressively in an unsupervised fashion from the data with pseudo labels, reducing the exhausting efforts of data labeling that requires expert knowledge. We have used the realtime data of Sentinel-2 for training the model and mapping the burnt regions. The obtained F1-Score of 0.87 demonstrates the effectiveness of the proposed model.

This paper introduces a machine learning-based approach for detecting electric poles, an essential part of power grid maintenance. With the increasing popularity of deep learning, several such approaches have been proposed for electric pole detection. However, most of these approaches are supervised, requiring a large amount of labeled data, which is time-consuming and labor-intensive. Unsupervised deep learning approaches have the potential to overcome the need for huge amounts of training data. This paper presents an unsupervised deep learning framework for utility pole detection. The framework combines Convolutional Neural Network (CNN) and clustering algorithms with a selection operation. The CNN architecture for extracting meaningful features from aerial imagery, a clustering algorithm for generating pseudo labels for the resulting features, and a selection operation to filter out reliable samples to fine-tune the CNN architecture further. The fine-tuned version then replaces the initial CNN model, thus improving the framework, and we iteratively repeat this process so that the model learns the prominent patterns in the data progressively. The presented framework is trained and tested on a small dataset of utility poles provided by “Mention Fuvex” (a Spanish company utilizing long-range drones for power line inspection). Our extensive experimentation demonstrates the progressive learning behavior of the proposed method and results in promising classification scores with significance test having p−value<0.00005 on the utility pole dataset.

In many real-world applications of computer vision complex domains, such as medical diagnostics and document analysis, the lack of labeled data often limits the effectiveness of traditional deep learning models. This study addresses these challenges by enhancing Unsupervised Curriculum Learning (UCL), a deep learning framework that automatically discovers meaningful patterns without the need for labeled data. Originally designed for remote sensing imagery, UCL has been expanded in this work to improve classification performance in a variety of domain-specific applications. UCL integrates a convolutional neural network, clustering algorithms, and selection techniques to classify images unsupervised. We introduce key improvements, such as spectral clustering, outlier detection, and dimensionality reduction, to boost the framework’s accuracy. Experimental results demonstrate significant performance gains, with F1-scores increasing from 68% to 94% on a three-class subset of the CIFAR-10 dataset and from 68% to 75% on a five-class subset. The updated UCL also achieved F1-scores of 85% in medical diagnosis, 82% in scene recognition, and 62% in historical document classification. These findings underscore the potential of UCL in complex real-world applications and point to areas where further advancements are needed to maximize its utility across diverse fields.

The effective training of supervised deep learning models requires the labeling of extensive datasets, a process that is often costly and labor-intensive. Such models also face significant challenges with overfitting on the training data with true labels, leading to suboptimal performance on new datasets with slight variations in capturing sources or regions. This paper introduces Multi-class Unsupervised Curriculum Learning (Multi-class UCL), a novel deep learning framework. We demonstrate the effectiveness of this framework on the  case study of land use and cover classification that bypasses the need for labeled data, thereby improving adaptability across different datasets. Multi-class UCL leverages pseudo-labels generated from a clustering technique to train the model and incorporates a selection process that ensures an equal representation of samples from each cluster, addressing the issue of class imbalance. The study evaluates the effectiveness of Multi-class UCL through comprehensive experiments on four diverse publicly available datasets: EuroSAT, SAT-6, RSSCN7, and UCMerced. These datasets have varying resolutions, come from different capturing sources, and encompass different geographical areas.The results demonstrate that the framework effectively learns and generalizes important features from the data, showing superior adaptability and performance across various datasets compared to traditional supervised models.