Abstract

Abstract
Humans spend most of their time indoors, where air quality and comfort are crucial to health and well-being. Elevated CO2 levels in buildings can reduce cognitive function, discomfort, and health issues. Indoor CO2 monitoring has emerged as a key focus in the literature, particularly in residential buildings, as it can play a vital role in helping to maintain adequate ventilation rates. The growing smart home market demands seamless integration and control, which are essential for implementing IAQ sensing devices. However, interoperability barriers between platforms and devices continue to hinder smart home adoption. To address these challenges, Matter protocol is starting to appear in the market. In this work, a wireless CO2 sensor is developed based on ESP32-C6 and SCD40 and integrated into a created Matter-enabled ecosystem formed with the Home Assistant open-source platform. The utilized hardware and software enable the usage of two different wireless communication technologies, WiFi and Thread, enhancing compatibility. The study highlights the rapid and seamless onboarding of the developed CO2 monitoring device into smart home ecosystems using the Matter protocol. As a result, once the device is successfully added to the ecosystem, the measurements can be accessed and analyzed through a mobile application, forming an IoT environment.

Abstract
Continuous Integration and Continuous Deployment are known methodologies for software development that increase the overall quality of the development process. Several robotic software repositories make use of CI/CD tools as an aid to development. However, very few CI pipelines take advantage of using cloud computing to run simulations. Here, a CI pipeline is proposed that takes advantage of such features, applied to the development of ATOM, a ROS-based application capable of carrying out the calibration of generalized robotic systems. The proposed pipeline uses GitHub Actions as a CI/CD engine, AWS RoboMaker as a service for running simulations on the cloud and Rigel as a tool to both containerize ATOM and execute the tests. In addition, a static analysis and unit testing component is implemented with the use of Codacy. The creation of the pipeline was successful, and it was concluded that it constitutes a valuable tool for the development of ATOM and a blueprint for the creation of similar pipelines for other robotic systems.

Abstract
Solar photovoltaic technology has grown significantly as a renewable energy, with unmanned aerial vehicles equipped with thermal infrared cameras effectively inspecting solar panels. However, long-distance capture and low-resolution infrared cameras make the targets small, complicating feature extraction. Additionally, the large number of normal photovoltaic modules results in a significant imbalance in the dataset. Furthermore, limited computing resources on unmanned aerial vehicles further challenge real-time fault classification. These factors limit the performance of current fault classification systems for solar panels. The multi-scale and multi-branch Reparameterization of convolutional neural networks can improve model performance while reducing computational demands at the deployment stage, making them suitable for practical applications. This study proposes an efficient framework based on reparameterization for infrared solar panel fault classification. We propose a Proportional Balanced Weight asymmetric loss function to address the class imbalance and employ multi-branch, multi-scale convolutional kernels for extracting tiny features from low-resolution images. The designed models were trained with Exponential Moving Average for better performance and reparameterized for efficient deployment. We evaluated the designed models using the Infrared Solar Module dataset. The proposed framework achieved an accuracy of 83.8% for the 12-Class classification task and 74.0% for the 11-Class task, both without data augmentation to enhance generalization. The accuracy improvements of up to 16.4% and F1-Score gains of up to 18.7%. Additionally, we achieved an inference speed that is 3.4 times faster than the training speed, while maintaining high fault classification performance.

Abstract
Chest radiography (CXR) is one of the most used medical imaging modalities. Nevertheless, the interpretation of CXR images is time-consuming and subject to variability. As such, automated systems for pathology detection have been proposed and promising results have been obtained, particularly using deep learning. However, these tools suffer from poor explainability, which represents a major hurdle for their adoption in clinical practice. One proposed explainability method in CXR is through contrastive examples, i.e. by showing an alternative version of the CXR except without the lesion being investigated. While image-level normal/healthy image synthesis has been explored in literature, normal patch synthesis via inpainting has received little attention. In this work, a method to synthesize contrastive examples in CXR based on local synthesis of normal CXR patches is proposed. Based on a contextual attention inpainting network (CAttNet), an anatomically-guided inpainting network (AnaCAttNet) is proposed that leverages anatomical information of the original CXR through segmentation to guide the inpainting for a more realistic reconstruction. A quantitative evaluation of the inpainting is performed, showing that AnaCAttNet outperforms CAttNet (FID of 0.0125 and 0.0132 respectively). Qualitative evaluation by three readers also showed that AnaCAttNet delivers superior reconstruction quality and anatomical realism. In conclusion, the proposed anatomical segmentation module for inpainting is shown to improve inpainting performance.

Abstract
This study reviews deep learning techniques and high-resolution satellite images to analyse urban morphology changes in developing countries. The goal is to create a system that can automatically identify and monitor changes in urban areas, such as buildings, roads, and green spaces, to provide accurate data for urban analysis and planning. The project aims to achieve detailed segmentation of urban objects in satellite images by utilising advanced convolutional neural network architectures and efficient image processing methodologies. The results from this study are expected to enhance urban planning and management, addressing the challenges faced by rapidly growing urban centres in developing countries. © 2025 The Author(s).