smartbuildingmag.com
07
'20
Written on Modified on
Qualcomm News
QUALCOMM BRINGS ADVANCED ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING CAPABILITIES TO ADDRESS MULTIPLE TIERS OF SMART CAMERAS WITH NEW SYSTEM-ON-CHIPS
Designed to deliver improved AI performance, multiple connectivity options, and design development efficiency while enabling more affordable devices.
Qualcomm Technologies, Inc., a subsidiary of Qualcomm Incorporated, today announced the introduction of the Qualcomm® QCS610 and Qualcomm® QCS410 system-on-chips (SoCs) to the Qualcomm® Vision Intelligence Platform. The QCS610 and QCS410 are designed to bring premium camera technology, including powerful artificial intelligence and machine learning features formerly only available to high end devices, into mid-tier camera segments. This comes at a time when intelligence at the wireless edge and robust connectivity are increasingly becoming the bar to overcome for smart camera applications in smart cities, commercial and enterprise, homes, and vehicles.
“The rise of the IoT and the associated growth in connected devices has created a shift toward the wireless edge. As we support the combination of on-device processing and multiple connectivity options for fast decision making and critical data transfer, AI becomes an even more transformative experience for businesses,” said Jeffery Torrance, vice president, business development, Qualcomm Technologies, Inc. “With the launch of QCS610 and QCS410, we are addressing increased customer demand for more integrated capabilities and improved AI-features with a variety of connectivity options – all in a cost-effective solution. This will allow our customers to put these exciting new capabilities to work in products from a wider range of ecosystem players.”
The QCS610 and QCS410 bring a highly integrated solution designed with multiple features to deliver a one-stop shop for our customers building camera-based devices. The new platform is built with our upgraded Qualcomm® KryoTM CPU, Qualcomm® AdrenoTM GPU and Qualcomm® HexagonTM DSP and includes our Qualcomm® Artificial Intelligence (AI) Engine designed to deliver up to 50% improved AI performance than the previous generation. This latest generation was rearchitected to deliver improved efficiencies and faster inferencing in the DSP resulting in more computing power and AI inferencing at the device level. Keeping the key workloads on the device can provide privacy and significantly reduce latency for the best user experience.
The Qualcomm Vision Intelligence Platform supports Linux and Android OS for a variety of IoT segments, including camera, Edge AI box, retail and robotics. Further enhanced capabilities include support for Microsoft Azure Machine Learning and Azure services. Dual ISPs support Video capture, Integrated audio, GNSS, hardware-based security and Qualcomm Technologies’ multiple connectivity options including 5G/4G, Wi-Fi, Bluetooth and Ethernet make this one of the most robust and advanced set of features available for non-premium smart cameras in a single SoC.
Some of the key features supported:
- Optimized Heterogenous Computing Architecture: Custom CPU, GPU and DSP design provides powerful compute capability that is engineered specifically for camera applications designed to utilize intensive processing while consuming less power.
- Superior Image Processing Support for dual 14-bit Qualcomm Spectra 230 ISP.
- Optimized AI Software Deployment: The Qualcomm® Neural Processing SDK completes the AI portfolio by provisioning a virtually seamless path for NN deployment. The modular tool supports varied frameworks like Caffe/Caffe2, TensorFlow/TensorFlow Lite and ONNX and performs optimized execution by utilizing the heterogeneous architecture to achieve desired performance.
- Modular Software and Support for Open Source Frameworks: Modular Linux software that can enable customizations including flexibility to build an optimal SW footprint and also extensive support for open source multimedia frameworks (GStreamer, Pulse-Audio and Wayland/Weston etc.,) and AI/ML framework (such as TF-Lite).
- Application Specific AI at the Edge Solutions: Work with an ecosystem of specialized companies to develop DNN solutions for specific AI use cases including face detection, face recognition, object tracking and people counting. Developers can take advantage of the Qualcomm® Neural Processing SDK for their custom network deployment.
