Land Use and Land Cover (LULC) classification using remote sensing data is a challenging problem that has evolved with the update and launch of new satellites in orbit. As new satellites are launched with higher spatial and spectral resolution and shorter revisit times, LULC classification has evolved to take advantage of these improvements. However, these advancements also bring new challenges, such as the need for more sophisticated algorithms to process the increased volume and complexity of data. In recent years, deep learning techniques, such as convolutional neural networks (CNNs), have shown promising results in this area. Training deep learning models with complex architectures require cutting-edge hardware, which can be expensive and not accessible to everyone. In this study, a simple CNN based on the LeNet architecture is proposed to perform LULC classification over Sentinel-2 images. Simple CNNs such as LeNet require less computational resources compared to more-complex architectures. A total of 11 LULC classes were used for training and validating the model, which were then used for classifying the sub-basins. The analysis showed that the proposed CNN achieved an Overall Accuracy of 96.51% with a kappa coefficient of 0.962 in the validation data, outperforming traditional machine learning methods such as Random Forest, Support Vector Machine and Artificial Neural Networks, as well as state-of-the-art complex deep learning methods such as ResNet, DenseNet and EfficientNet. Moreover, despite being trained in over seven million images, it took five h to train, demonstrating that our simple CNN architecture is only effective but is also efficient.

Socio-ecological dynamics affect the ecosystem services supply and are relevant to generate effective water management strategies; this condition is considered to evaluate under a holistic approach, the water ecosystem services (WES) in an Andean supply basin (ASB) in Colombia. This analysis focus on the connection of biophysical and sociocultural components for the multi-purpose use of water based on The Soil and Water Assessment Tool (SWAT) modelling for Las Piedras River Basin (LPRB). The generated Hydrological Response Units (HRUs), allows to estimate the capacity of the basin for supplying water (quantity) in adequate conditions (quality) for local populations in rural and urban areas, as well as WES zoning. The model was calibrated and validated to generate a baseline scenario, which was complemented with social cartography and participative workshops. The results indicate a low concentration of nitrogen and phosphorus, boosted by specific agro-ecological strategies developed by local communities; however, there are health risks for populations downstream and those that are supplied with water directly from the source. Additionally, Land Use and Land Cover (LULC) affects water availability, which demands restoration and conservation strategies to maintain WES supply for socioeconomic and cultural purposes, since different views on the available WES converge in the basin.

Academic training of professionals, especially in engineering fields, should enable them to reliably interpret their surroundings for informed decision-making. This is crucial for engineering students who continuously quantify and anal-yze system operations. In the current context, where in-person and virtual work blend together thanks to technological advances, professional education must establish flexible alternatives to assimilate phenomenological knowledge without sacrificing essential aspects of real experimentation. Remote laboratories emerge as an alternative to strengthen these aspects through distance learning. This paper presents a proposed and developed functional architecture scheme for remote interaction-based mechanical physics laboratories, allowing the configuration of initial operating conditions, execution of experiments, and data collection for three fundamental physics practices: free fall, Hooke’s law, and parabolic motion. The proposed architecture is scalable under the Internet of Things approach, using the necessary hardware and software elements with user-friendly interaction interfaces, making the developed remote laboratories a viable option for engineering students’ education, without requiring physical presence for their execution.

La seguridad ha sido uno de los puntos claves en el diseño vehicular, por lo que uno de los principales objetivos es implementar sistemas de alerta para notificar al conductor sobre algún proceso inadecuado o atípico en su conducción, con el fin de evitar accidentes que afecten a sus ocupantes, así como a terceros; un ejemplo, de esto se observa en el auge de los vehículos autónomos. De acuerdo con la Organización Mundial de la Salud, en el 2016 se presentaron 1.35 millones de muertes relacionadas con accidentes de tráfico, por ello, actualmente se crean más sistemas para monitorizar el ambiente alrededor del vehículo de modo que se garantice una conducción segura en todo momento. Esta investigación presenta el primer sistema de asistencia a la conducción desarrollado para Colombia, el sistema detecta y reconoce señales de tránsito preventivas y reglamentarias basado en clasificadores Haar, lo cual permite que su precisión no se afecte debido a las rotaciones y escala de las señales presentes en un viaje sobre un trayecto real. El sistema reconoce salidas de carril, estimación de la dirección de la curva y detección de obstáculos que sobresalen en la carretera utilizando algoritmos de visión por computadora convirtiéndolo en un sistema de bajo costo computacional. Además, esta investigación proporciona los primeros clasificadores en cascada resultantes para la detección de señales reglamentarias y preventivas colombianas. El sistema es probado en ambientes reales de carreteras colombianas obteniendo una precisión superior al 90%. La investigación demuestra que métodos basados en visión por computadora son competitivos frente a propuestas actuales como las redes neuronales profundas.