Embedded World 2018: IoT adds pressure to teach machines to 'see'


Visual input is arguably the richest source of sensor information. Scientists and engineers have been trying to understand and exploit imaging technologies for many decades now, developing algorithms for vision applications that enable computing machines to'see'.

New Software Library For Vision Intelligence In Mobile

AITopics Original Links

A new computer vision (CV) software library has been launched for the development of "vision-enabled" applications targeting the mobile, home, PC, and automotive markets. Built by CEVA Inc, the new library is optimized for the firm's own CEVA-MM3101 imaging and vision platform. This is positioned as a tool for application developers to add vision capabilities to System-on-Chip (SoC) systems incorporating the CEVA-MM3101. NOTE: So-called "vision enabled" applications can be found in areas such as wireless (or wired) sensor networks, mobile computing scanning apps, PCs, smart TVs, natural user interface (NUI) devices, and advanced driver assistance systems (ADAS). CEVA-CV is based on OpenCV, a standard library of programming functions for computer vision processing.

Irida Labs' NoiseSweeper and EnLight Software Now Available on Cadence Tensilica Imaging/Vision DSPs


In addition, IRIS-NoiseSweeper software has also been ported to the Tensilica Vision DSP to provide premium video quality and cleaner still-images. The software is targeted at image noise reduction for high-resolution still images. The technology features an automated noise profile estimation that reduces calibration needs and shortens development time. It operates in a single frame and eliminates any motion-blurring phenomena associated with multi-frame techniques. The Tensilica family of imaging/vision DSPs was designed for the complex algorithms in imaging, video and computer vision applications including innovative multi-frame image capture, video pre- and post-processing, object and face recognition, low-light enhancement and many other complex tasks.

Press Releases


CEVA, Inc. (NASDAQ: CEVA), the leading licensor of signal processing IP for smarter, connected devices, today announced that Novatek Microelectronics, Taiwan's 2nd largest fabless IC design house, has licensed and deployed the CEVA-XM4 intelligent vision DSP for its next-generation vision-enabled System-on-Chips (SoCs) targeting a range of end markets requiring advanced visual intelligence capabilities. Novatek's current camera SoC lineup for car DVR and surveillance systems integrates the 3rd generation CEVA-MM3101 imaging & vision DSP and is shipping in volume. By integrating CEVA-XM4 as a dedicated vision processor in their next-generation SoC designs, Novatek and its customers can rapidly deploy highly-sophisticated vision algorithms to enable advanced applications such as surveillance systems with face detection and authentication, drone anti-collision systems and advanced driver assistance systems (ADAS). These types of applications are built utilizing CEVA's Deep Neural Network (CDNN2), a proprietary software framework that enables deep learning tasks to run on the CEVA-XM4 and outperform any GPU or CPU-based system in terms of speed, power consumption and memory bandwidth requirements. "The CEVA-XM4 is an exceptional processor for imaging and computer vision, offering outstanding performance, flexibility and power efficiency for applications requiring visual intelligence capabilities," said Allen Lu, Assistant Vice President of iVoT SBU, Novatek.

A List of Chip/IP for Deep Learning


Machine Learning, especially Deep Learning technology is driving the evolution of artificial intelligence (AI). At the beginning, deep learning has primarily been a software play. Start from the year 2016, the need for more efficient hardware acceleration of AI/ML/DL was recognized in academia and industry. This year, we saw more and more players, including world's top semiconductor companies as well as a number of startups, even tech giants Google, have jumped into the race. I believe that it could be very interesting to look at them together. So, I build this list of AI/ML/DL ICs and IPs on Github and keep updating. If you have any suggestion or new information, please let me know. The companies and products in the list are organized into five categories as shown in the following table. Intel purchased Nervana Systems who was developing both a GPU/software approach in addition to their Nervana Engine ASIC. Intel is also planning in integrating into the Phi platform via a Knights Crest project.