![]() We will launch a road network competition this fall for the 3rd SpaceNet competition, and we will provide details regarding this competition soon. Going forward, we are planning new SpaceNet competitions beyond automated building footprint extraction. The dataset will continue to be available on SpaceNet’s AWS page and the solution code base will remain available on the SpaceNet Github repository. The classids for features can now be inferred using Mapbox GL filters, using code from Label Maker (thanks). ![]() We encourage readers to explore the winning solutions developed by XD_XD, Wleite, and Nofto. Generating training data using SpaceNet Vegas imagery, OSM vector tiles as a label source, and buffering to convert lines to polygons. While the competition submissions excel in locating building footprints in Las Vegas, the submissions struggle in Shanghai and Khartoum. The results from 2nd SpaceNet competition show that performance is regionally specific. The full write up and code can be found here. Nofto modified the code to use the multi-spectral pan-sharpened imagery and modified the random forest implementation. Nofto achieved third place in the competition using Wleite’s code from the first SpaceNet competition. You can read more about his solution here. Each polygon candidate is evaluated for its probability of having an IoU over 0.5, and polygons below a certain city-dependent cut-off are discarded. The brute force polygon matching is performed based on the edge detection and pixel classification.Ĥ. The first classifier determines whether a pixel belongs to a border and the second is whether a pixel is inside or outside a building.ģ. He uses random forests to create two binary classifiers. He applies edge detection to each band of the multiband image.Ģ. His implementation is a custom java application that does not leverage deep learning frameworks. Wleite achieved second place using his code base from the first competition while providing a more methodical evaluation of possible improvements. The red outlines are false positives algorithm proposals and the yellow outlines represent false negatives. ![]() The blue outline represents the ground truth, the green outlines are true positive algorithm proposals. Figure 3: Results from XD_XD’s implementation: From Top left, Clockwise (AOI 2 Vegas: Image 1014, AOI 2 Vegas: Image 104, AOI 5 Khartoum: Image 991, AOI 3 Paris Image 1720). ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |