Indoor scene recognition and semantic information can be helpful for social robots. Recently, in the field of indoor scene recognition, researchers have incorporated object-level information and shown improved performances. This paper demonstrates that scene recognition can be performed solely using object-level information in line with these advances. A state-of-the-art object detection model was trained to detect objects typically found in indoor environments and then used to detect objects in scene data. These predicted objects were then used as features to predict room categories. This paper successfully combines approaches conventionally used in computer vision (YOLO) and Term Frequency-Inverse Document Frequency (TF-IDF). These approaches could be further helpful in the field of embodied research and dynamic scene classification, which we elaborate on.

This paper investigates if Corporate Social Responsibility (CSR) reports published by a selected group of Nordic companies are aligned with the Global Reporting Initiative (GRI) standards. To achieve this goal, several natural language processing, and text mining techniques were implemented and tested. We extracted strings, corpus, and hybrid semantic similarities from the reports and evaluated the models through the intrinsic assessment methodology. A quantitative ranking score based on index matching was developed to complement the semantic valuation. The final results show that Latent Semantic Analysis (LSA) and Global Vectors for Word Representation (GloVE) are the best methods for our study. Our findings will open the door to the automatic evaluation of sustainability reports which could have a strong impact on the environment.

Generating early warning signals of reduced resilience in ecosystems is crucial for conservation and management endeavors. However, the practical implications of such early warning signal systems are still limited by the lack of data and uncertainties associated with predicting complex systems such as ecosystems. This research aims to investigate the feasibility of developing an early warning system capable of identifying an upcoming critical transition within mangrove forest ecosystems. Using time series analysis of remote sensing images, the resilience of mangrove forests was explored across two distinct study sites. One site (Qeshm Island) has been adversely affected by land-use and land-cover changes, while the other (Gabrik) serves as a reference ecosystem. The study uses data from the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite to quantify three remotely sensed indices: the Normalized Difference Vegetation Index (NDVI), the Modified Normalized Difference Water Index (MNDWI), and the Modified Vegetation Water Ratio (MVWR). In addition, Landsat data has been used to explore temporal alterations in land-use and land cover. To identify early warning signals, indicators such as autocorrelation (acf(1)), standard deviation (SD), and skewness (SK) are applied. The findings indicate a signal of reduced resilience by a significant increase in NDVI statistical metrics (acf(1): 0.50, SD: 0.9). Although MNDWI showed significant early warning signals in Qeshm Island (acf(1); 0.86, SD: 0.90), it provided a false alarm in the reference study site. MVWR failed to generate early warning signals of reduced resilience (acf(1); -0.100, SD: -0.07, SK: -0.21). Land-use land-cover change may explain reduced resilience in the forests. This study not only emphasizes the potential of remote sensing in monitoring the state of mangrove forests but also contributes to advancing our understanding of ecosystem dynamics. The findings of this study can be integrated with ecosystem management tools to enhance the effectiveness of conservation efforts aimed at safeguarding mangroves. This is the first report of the successful application of remote sensing in generating early warning signals of reduced resilience within mangrove forests in the Middle East.

Optimizing intensive care resources using predicting modeling is paramount for fighting the COVID-19 pandemic. In this paper, we model the admission of COVID-19 patients in intensive care units (ICU) in Colombia using openly available data gathered from 18 March 2020 to 14 October 2020. After an intensive preprocessing of the data, we trained four different machine learning models using four different strategies for handling the imbalanced features. Our findings show that our best model (XGBoost) effectively predicts an Area Under the Curve (AUC-ROC) of 0.94, in line with the state-of-the-art results obtained in other predictive models obtained with medical data.

This study aims to enhance multimodal transportation systems by mitigating CO2 emissions and enhancing operational efficiency. It introduces a novel centralized cargo concentration approach tailored for regions facing geographical challenges. The method proposes for direct transportation of cargo from its loading origin to export ports, bypassing intermediate hubs. The mathematical model determines the most efficient means of transport for each route, factoring in variables like distance, volume, and cargo type. Research results indicate that, in scenarios with a high concentration of cargo, multiple hubs may not be necessary, which could streamline transportation and logistics operations. Modal preferences vary based on regional dynamics and cargo characteristics, with rail and sea transport emerging as preferable options under specific circumstances, surpassing the efficacy of road transport. The proposed model showcases reductions in both logistics costs and CO2 emissions compared to road-centric scenarios. It underscores the strategic integration of diverse transportation modes as pivotal for enhancing efficiency and sustainability. This study furnishes a framework adaptable to optimizing multimodal transportation systems in regions sharing similar geographical and logistical attributes.