Open LeosonH opened 5 years ago
Public code is available, git sites given in #8.
We'd love to develop models which no longer work with static images, but with continual image input as our brains must deal with it. As of now, we're still thinking of ways to deal with this step up in complexity.
It will be interesting to experiment if the model will be amenable to more complex image analysis tasks - as mentioned in #8, if a public version of the model is available, it will be very useful!
Would be exciting to see if robust contextual learning is possible (as mentioned in #2), which I imagine would be very useful for improving the efficiency of key processes in remote sensing, OCR, UI/UX analysis, etc.