Leveraging dynamics in computer vision problems: user friendly and theoretically sound tools.
Many Computer Vision tasks deal with data, such as video or image sequences, that are highly temporally coherent in content. Some hallmark examples may include tracking, activity recognition, video segmentation, pose estimation . . . etc. These temporal relations in the data are called dynamics and...
| Published: |
|
|---|---|
| Online Access: | http://hdl.handle.net/2047/D20251058 |
| Summary: | Many Computer Vision tasks deal with data, such as video or image sequences, that are highly temporally coherent in content. Some hallmark examples may include tracking, activity recognition, video segmentation, pose estimation . . . etc. These temporal relations in the data are called dynamics and can be thought of the signature of a temporal signal. Dynamics can be ex- tremely useful in distinguishing competing hypothesis and solving difficult computer vision prob- lems.
However one needs proper set of tools to benefit from dynamics of the signal. |
|---|