Deep learning – bringing the benefits into low-power embedded systems

Researchers have been investigating ways to mimic the human brain in computers for many years. Imitating the human brain, where many neurons are connected to each other and able to learn over time to identify new patterns and recognise objects, was mostly challenging in the vision domain, where the amount of data being processed (and the performance required) is high. Recent advances in quality of vision algorithms- Deep neural networks, in conjunction with the significant increase in raw embedded computing horsepower have brought Deep Learning to the forefront of technology exploration in many areas and applications.

Join CEVA and Phi Algorithm Solutions vision experts to hear about:

•Overview to deep learning and how it may benefit embedded products.

•The development flow of deep neural networks from offline training onto low-power consumer devices and explore the various tradeoffs when developing for embedded system.

• How to deploy deep neural networks based applications quickly yet allowing enough flexibility for future market changes.