High resolution remote sensing’s Achilles heel: Linking image objects (this is a 'thing') and high-level semantics (what is this 'thing'?)

The spatial resolution of available satellite-based imagery has shrunk dramatically to just over 30 cm. However, this blessing is quickly becoming a curse because, at this point, increased spatial resolution is actually making it more difficult to extract meaningful information on ground targets. As image interpretation shifts from traditional remote sensing approaches to the realm of visual computing, we have to rethink the way in which we map pixels to objects (this collection of pixels is a ‘thing’) and give meaning to those objects (this ‘thing’ is a fur seal).