09 Aug Learning and Video
Learning and Video
We asked IT Support Woking – What are some potential applications of deep learning applied to video?
What We Know
Learning and Video are both very interesting in so many ways. It is also a good thing to know that we have actually made much stronger gains with the vision due to ImageNet at the human levels. We are so far away from understanding the human eye. There are many reasons due to which you are at this kind of stage. Limited amount of information can only be available in single snapshots that we may get in the images. There may be some lack of the large labelled dataset. The imageNet is something of great boon and is scaling up with so many categories of things that humans may find it hard for recognizing. Video has got potential for solving all these and can help machines get human vision.
A 3D View of the World
There are so many sequences of frames associated with each video and are related to each other and can provide a much richer perspective towards the real world that allows models for creating a 3D view of the world. There is no need for you to have stereo vision. There is also a continuity with the frames that can allow an unsupervised form of learning algorithms to the extract of the valuable image information without actually resorting to the labels.