- Published November 2017
15 expert presentations, over 190 registered participants and dozens of exhibitors – the first Embedded VISION Europe Conference on October 12th and 13th, organized by EMVA was very well received by the industry. Leading players including Intel, AMD, and Qualcomm gave presentations on the incredible image processing capabilities of their chips. From car and people tracking to face recognition – tasks that not long ago required large supercomputers can now be performed on compact, light-weight and low-power specialist processors.
When it comes to unsupervised machine learning, even of complex objects, different platforms are available showing impressive proficiencies that are often close to, or even beyond, human performance. Some are capable of performing both deep learning and recognition tasks on the same powerful GPU-like hardware. More compact chips with less power consumption rely on external deep learning of deep neural networks, e.g. training in the cloud, and then execute recognition tasks based on such learned models.
On the programming side of embedded vision systems, the need for lengthy development cycles has been eliminated in some cases; Mathworks showed a framework to develop advanced imaging algorithms in MATLAB, which can generate code for embedded processing platforms.
In addition to these insights into state-of-the-art embedded vision, the two half-days of the conference plus an evening reception offered lots of opportunities for networking among leading industry players. We hope for a further issue of this event format and are looking forward to attending again.
Active Silicon is your partner in Embedded Vision Systems for demanding applications in industrial manufacturing, medical, traffic, security, entertainment, and many other fields. We leverage the latest technologies that fulfill your requirements of reliability, performance, ruggedness and long-term availability. Please contact us to discuss your strategy or current project in Embedded Vision.