Bookmark and Share

General Discussion

Current EV challenges
2 replies [Last post]
Moon
Offline
Last seen: 6 years 41 weeks ago
Level 1: Prestidigitator
Joined: 2012-10-17
Points: 1

What are the current embedded vision challenges that can be a topic for a PHD thesis?

Many thanks in advance for your assistance.

squreshi
Offline
Last seen: 2 years 1 week ago
Level 4: Thaumaturgist
Joined: 2011-05-31
Points: 95

There are so many I don't know where to start, but some of the currently "en vogue" active topics include (but are not limited to):

  1. gesture recognition
  2. high-dynamic range imaging
  3. super-resolution
  4. object tracking
  5. multi-view geometry
  6. face recognition

Any embedded embodiment of the above must take into account the constrained resources available to most embedded devices. So the implementation must be geared towards reduced computational requirements, power requirements, and more than likely both.

Amit Shoham
Offline
Last seen: 15 weeks 3 days ago
Level 4: Thaumaturgist
Joined: 2011-05-28
Points: 60

Those are all interesting challenges from the algorithm/application perspective.

From a hardware perspective, there's a "big picture" challenge:  given the wide range of vision applications, and the fact that most of them require various accelerators (e.g. DSP, GPU, custome video/image processing engines), how can silicon vendors best address the needs of embedded vision applications?