因此，即使在这项任务上花费了数千小时，谷歌的员工仍然不能成功地绘制出遍布全球2000万英里的道路的大部分。 幸运的是，MIT计算机科学和人工智能实验室（CSAIL）的研究人员可能已经找到了这个问题的解决方案。他们开发了一套道路识别系统Roadtracer，能够直接基于航拍卫星图精准地识别公路、街道和小径等，绘制城市地图。 这套识别系统基于神经网络，以AI步进式识别像素点是否是路径，尽管在公园和林地的识别精度不高，但也能够解决无实地道路数据的情况下，绘制城市地图的难题。该系统可能特别适合映射世界上地图经常过时的部分地区，如发展中国家的偏远地区和农村地区。 ...
So, even if has spent thousands of hours on this task, Google employees still can't successfully mapped the most in all over the world, 20 million miles of roads. Fortunately, MIT's computer science and artificial intelligence laboratory (CSAIL), researchers may have found the solution to this problem. They developed a set of road recognition system Roadtracer, aerial can directly based on the satellite image accurately identify roads, streets and paths, such as map city. The recognition system based on neural network to the AI step by step to identify whether pixels is path, while in the park and woodlands of the identification accuracy is not high, but also can be solved without the road data, map of urban problems. The system may be particularly suitable for mapping the world map often outdated parts, such as remote areas and rural areas of developing countries. ...
