Deep-Feature-Flow
Deep-Feature-Flow copied to clipboard
For real-time object detection purposes
Would this be suitable for real-time object detection, say for autonomous vehicle purposes?
My concern is that when a new object appears, will it only be detected when the next keyframe is processed? If so then, there will be a delay before an object is detected, which is undesirable for real-time object detection purposes.
If not, does the model handle the appearance of new objects between keyframes somehow?