Pollution in the form of litter in the natural environment is one of the great challenges of our times. Automated litter detection can help assess waste occurrences in the environment. Different machine learning solutions have been explored to develop litter detection tools, thereby supporting research, citizen science, and volunteer clean-up initiatives. However, to the best of our knowledge, no work has investigated the performance of state-of-the-art deep learning object detection approaches in the context of litter detection. In particular, no studies have focused on the assessment of those methods aiming their use in devices with low processing capabilities, e.g., mobile phones, typically employed in citizen science activities. In this paper, we fill this literature gap. We performed a comparative study involving state-of-the-art CNN architectures (e.g., Faster RCNN, Mask-RCNN, EfficientDet, RetinaNet and YOLO-v5), two litter image datasets and a smartphone. We also introduce a new dataset for litter detection, named PlastOPol, composed of 2418 images and 5300 annotations. The experimental results demonstrate that object detectors based on the YOLO family are promising for the construction of litter detection solutions, with superior performance in terms of detection accuracy, processing time, and memory footprint.

The aquaculture industry is a main industry in Norway, and it must be sustainable, i.e. experience long-term growth and development: It is necessary to build an environmentally sustainable aquaculture industry that minimizes risks to the marine environment and biological diversity, – including the transmission of fish diseases. The process of fish disease transmission in aquaculture systems is influenced by many factors, including individual (fish and pathogen) conditions, movement behavior and environmental conditions. Fish disease dynamics originates from a complex system, and the transmission of viruses is an unstable process, making it difficult to predict and analyze. In preparation for this paper, we built an agent-based model to predict patterns of pathogen transmission with the purpose of identifying risks and hazards in the space and time domains. The model presented explores the potential effects of different factors, such as the conditions of agents, movement behavior and environmental conditions, on the simulated spread of a fish disease. We applied the model developed to different case studies in the Norwegian fjords. The results demonstrated how the infection risk at any point around the infected site is dependent on both the pathogen and the fish density at that point, and the infection risk increases when the pathogen or fish densities increase. The pathogen density decreases exponentially as a function of an increase in the water temperature, and the pathogen density increases with the velocity of the current or the fish density at the infected site. The pathogens are moved faster by higher current velocity, so this will slow the infection process at the local infected site. Nevertheless, the current will carry the pathogens to nearby places faster. The direction of the current is very important since the pathogens are predominantly moved by the currents. The agent-based method helps us advance our understanding of pathogen transmission and builds risk maps to help us reduce the spread of infectious fish diseases. By using this method, we may study the spatial and dynamic aspects of the spread of infections and address the stochastic nature of the infection process.

Transforming the livability of citizens by combining data, resources, and infrastructure is one of the fundamental goals of a smart city. A smart city is an aggregation of data and live analysis. Smart mobility can be equated with the smart planning of vehicle infrastructure in smart cities. Digital twin (DT) is an advanced concept for creating digital replicas of physical assets in a virtual space to mimic the original assets for different test scenarios. With DTs, efficient smart mobility can be planned in smart cities to improve transport systems. However, manually generating digital replicas in a virtual environment is cumbersome. Automatic 3D reconstruction of vehicles to be used as DTs in a virtual environment can save a lot of labor. Therefore, we propose a digital twin authoring tool (DTAT) for generating DTs from a smart mobility perspective. The proposed DTAT ingests diverse data types from internet-of-things sensors, such as images and 3D CAD models. The reconstructed 3D models are used as DTs of physical vehicles in a virtual environment. The 3D reconstruction in our proposed DTAT is based on pre-trained models. To provide an immersive experience to city stakeholders, we develop a virtual reality (VR) based representation of vehicles in cyberspace. Our approach extends beyond mere visualization, enabling the running of various simulations within the virtual environment to gain comprehensive analytical insights. These simulations include advanced applications like public transport system optimization, using 3D models and real-time data to simulate and improve vehicle routes, minimizing travel times and maximizing coverage. This particular application demonstrates the practicality and effectiveness of our DTAT in enhancing urban mobility management. We showcase the capabilities of our proposed solution through a VR application, illustrating how it can be a transformative tool in the planning and management of smart city transport systems. Implementing our DTAT streamlines the creation of digital twins and significantly contributes to the smarter, more efficient, and sustainable movement within urban landscapes.