The conservation of marine resources requires constant monitoring of the underwater environment by researchers. For this purpose, visual automated monitoring systems are of great interest, especially those that can describe the environment using semantic segmentation based on deep learning. Although they have been successfully used in several applications, such as biomedical ones, obtaining optimal results in underwater environments is still a challenge due to the heterogeneity of water and lighting conditions, and the scarcity of labeled datasets. Even more, the existing deep learning techniques oriented to semantic segmentation only provide low resolution results, lacking the enough spatial details for a high performance monitoring. To address these challenges, a combined loss function based on the active contour theory and level set methods is proposed to refine the spatial segmentation resolution and quality. To evaluate the method, a new underwater dataset with pixel annotations for three classes (fish, seafloor, and water) was created using images from publicly accessible datasets like SUIM, RockFish, and DeepFish. The performance of architectures of convolutional neural networks (CNNs), such as UNet and DeepLabV3+, trained with different loss functions (cross entropy, dice, and active contours) was compared, finding that the proposed combined loss function improved the segmentation results by around 3%, both in the metric Intercept Over Union (IoU) as in Hausdorff Distance (HD).
The combination of light detection and ranging (LiDAR) sensors and unmanned aerial vehicle (UAV) platforms have garnered considerable interest in recent years because of the wide range of applications performed through the generation of point clouds, such as surveying, building layouts and infrastructure inspection. The attributed benefits include a shorter execution time and higher accuracy when surveying and georeferencing infrastructure and building projects. This study seeks to develop, integrate and use a LiDAR sensor system implemented in a UAV to collect topography data and propose a procedure for obtaining a georeferenced point cloud that can be configured according to the user’s needs. A structure was designed and built to mount the LiDAR system components to the UAV. Survey tests were performed to determine the system’s accuracy. An open-source ROS package was used to acquire data and generate point clouds. The results were compared against a photogrammetric survey, denoting a mean squared error of 17.1 cm in survey measurement reliability and 76.6 cm in georeferencing reliability. Therefore, the developed system can be used to reconstruct extensive topographic environments and large-scale infrastructure in which a presentation scale of 1/2000 or more is required, due to the accuracy obtained in the work presented.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.