And that is creating 3D geometry by looking at a 2D color image.

那就是用2D彩色图像来创建3D几何结构。

The teacher is a standard pose estimation neural network that looks at a color image and predicts the pose of the humans therein.

老师就是一个标准的姿势预测神经网络，它通过观察彩色图像预测其中的人物姿势。

In color images, every pixel is represented by three binary sequences that correspond to the primary colors.

在彩色图像中，每个像素都由对应于原色的三个二进制序列表示。

The color images include the Robinson projection, Mercator projection, azimuth projection, and orthographic projection of the Eastern and Western hemispheres of the Red Planet, with a spatial resolution of 76 meters.

本次发布的影像图为彩色，包括按照制图标准分别制作的火星东西半球正射投影图、鲁宾逊投影图和墨卡托投影加方位投影图，空间分辨率为 76 米。

This is not much of a problem with color images, because we can clearly see everyone in an image, but the radio signals are more difficult to read when they reflect off of multiple bodies in the scene.

这对彩色图像来说不是什么大问题，因为我们可以看清每个人，但无线信号经过多次反射之后更难分辨场景中的人物。