 Thank you for introduction. So today, my presentation will be about the solid solution for UIV point clouds, how raw point clouds can be classified, and possible examples of end products that you can export from that classified point cloud. Why I choose to concentrate on data collected by UIVs. So launch of affordable LiDAR systems, I think, changed market a lot, and LiDAR system become more affordable. As well, spatial data collection before was more like a large-scale project handled by national mapping agencies. But the drone can be deployed easily in any areas. And then, basically, the start of data collection is quite easy. As well, multi-sensor integration is easier. And besides LiDAR data, you can collect images, RGB, thermal, multi-spectral. And as well, Thera-solid did quite a lot development for UIV data processing and simplified the workflow quite a lot. So this is what I want to talk in this presentation. Who knows already Thera-solid software? OK, there is some hands. So Thera-solid is the software for point cloud and images processing. A bit about the company. We are in Finland, the headquarters. And we have users all over the world to have the idea how many tool sets we have. So we calculate that it is for eight industries. And what kind of data you can process. So any point clouds, no matter the source, sensor, and the scale of the project. I will concentrate today on data captured by UIVs. But basically, any data as handheld, collected, or mobile data, as well for the geometric point clouds, also possible. Classified and vectorized data can be used in planning, construction sites, maintenance. And in this example, you can see how different attributes can be used for data classification. For example, dimension, echo returns, then also elevation from ground, intensity value. The paint markings are detected using intensity value. Cars classified using grouping method or segmentation. As well, traffic is removed using moving objects routine. Then slope can be used for rats and potholes detection. The triangulated model can be helpful to vectorize, for example, tram lines. As in this example, classified buildings can be as an input for 3D building models creation. So software can automatically vectorize those. And later, tram lines, buildings, detected trees can be used in further modeling. So basically, you need to classify the raw point cloud in order to use it as a real-world representation in further models. Now, about workflows in TeraSolid. So what are the possibilities for classification? First option is to use wizard. So basically, the input of the data and classification is handled by wizard. It is simplified, so you don't need to set up many things. The second scenario allows you to change the settings, modify the macros according to your data set or project needs. And the workflow stays still automatically. So as a step one wizard data import, there's quite many sensor choices. And by this, software knows what kind of data to expect, so how settings can be adjusted according to that in later classification tasks. As well, this wizard not just import the data but can reproject to local coordinate systems or adjust to geoid model. And then the second step combines multiple classification steps in one dialogue. You can run either everything at once or with the manual inspection between the steps. So first of all, this wizard removes the noise, cut the overlap, improve the matching, then it moves to ground classification, and then above ground features. Here you can see the journey from raw point cloud classification to classified point cloud. So in the beginning, the data is just imported. You can see a lot of overlap between the lines, and data is still quite noisy. In this one, the overlap is cut out, some smoothing as well applied. Ground classification combines multiple steps. So in the beginning, it tries to detect potential ground surface and done from that to collect some points as a final ground classification. Then layering data by height from ground and wizard classification. So automatically wizard detected trees, vegetation, houses, so roofs, walls, and as well cars. From this point, you can go further with some manual classification improvement or export already ready results, as for example, tree locations. Because during the tree classification, each tree was identified. So you can export the location of the tree, and as well some dimensions attributes. Classified roofs can be used as an input for 3D building models. And as I said, so from here, you need to continue either with industry-specific tools or continue with the results. And one of the example is terrain model. So when you have ready the ground surface, either you need to improve that with matching or adjustment to the ground control points, or you can go straight to triangulate that model production. As well, it can be just simple digital elevation model from ground points, or you can add break lines next to the river or roads. And contours production as well as possible. Over, you can calculate the volume of the stockpiles. This is more important maybe for mining business. And then another example is power line maintenance projects, where raw point classification goes further to wires and towers classification. As well, one of the goals is to detect the wires. So then you can use those wires to detect dangerous vegetation buildings in the clearance zone. When vectorization is happening, also the software classifies the points along the wires. And then for the dangerous vegetation, for example, falling tree logic can be used. So this ends my presentation. To sum up, Therosolet has a solution for UAV data processing. And with the classified point cloud, you can go further industry specific with industry specific requirements or already export required results. Thank you for listening.