Road and Building Detection

Globally available land cover maps derived from Planetscope imagery classifying pixels as road, building or neither. Generated on a weekly or monthly basis, <a href="https://www.planet.com/products/analytics/" target="_blank"> Planet&#39;s Road &amp; Building Detection </a> can be used to stay up to date with the latest development around the globe.<br>By leveraging a semantic segmentation model on all the PlanetScope imagery published in a given area for a given week or month, we assign a probability value (0-255) that the pixel represents a road, building or neither object. These results are aggregated together into a weekly or monthly layer and this process is repeated through time, generating a time-series of pixel values ready for analysis.

全球范围内可用的土地覆盖图，由 Planetscope 图像生成，将像素分类为道路、建筑或非此类。该数据集以每周或每月为周期更新，<a href="https://www.planet.com/products/analytics/" target="_blank"> Planet 的道路与建筑检测产品</a>，旨在帮助用户及时了解全球范围内的最新发展动态。通过在特定区域、特定周或月发布的所有 PlanetScope 图像上应用语义分割模型，我们为像素分配一个概率值（0-255），以表示该像素代表道路、建筑或非此类物体。这些结果被汇总成每周或每月的图层，并且这一过程随时间重复进行，从而生成可供分析的像素值时间序列。