Connected smart cameras have better image recognition

18th October 2016
Source: Qualcomm
Posted By : Alice Matthews
Connected smart cameras have better image recognition

The subsidiary of Qualcomm Incorporated, Qualcomm Technologies, has expanded its portfolio of technologies to support a generation of video and imaging devices. The Qualcomm Snapdragon connected camera portfolio includes a comprehensive suite of camera platforms featuring enhanced hardware and software capabilities.

These provide on-camera deep learning and video analytics, 4K HEVC video encoding, support for the Linux operating system, broad connectivity options including 4G LTE, WiFi, Bluetooth and advanced features for robust hardware-based security.

Snapdragon connected camera platforms are built on years of experience and technology utilised in billions of cameras in the mobile space, and are designed to help manufacturers accelerate development of smart, connected cameras in varied form factors including 360/VR, sports, bodycams, dashcams and IP security cameras. Updates to the platforms include:

Snapdragon 625 processor
Built on a 14nm process, the Snapdragon 625 is designed to use up to 25% less power compared to its predecessor in the camera chipset portfolio, while also designed to deliver 40% more CPU DMIPs processing with GPU and DSP for advanced imaging and deep learning processing. Already proven in smartphones, the processor features dual image signal processors for up to 24MP images and 4K video at 30fps HEVC capture, octa-core 64-bit CPU, an integrated X9 LTE modem for peak speeds up to 300Mbps download/100Mbps upload, and support for WiFi 802.11ac, Bluetooth and GNSS/location.

These connectivity options are designed to allow for real time streaming to social media platforms such as Facebook Live, YouTube, Twitter, Periscope and iQiyi. The integrated Qualcomm Adreno 506 GPU is engineered to allow VR/360 cameras to process fisheye dewarping and stitching using inputs from multiple image sensors. This is designed to provide a richer and more consistent user experience compared to traditional panoramic photos using smartphones. The Snapdragon 625 also is designed to provide advanced hardware-based security features including secure boot, cryptographic hardware engines, trusted execution environment, and storage with rich security capabilities and secure key provisioning.

Linux support
Qualcomm Technologies' connected camera roadmap now supports Linux in addition to Android, designed to help accelerate time-to-commercialisation for manufacturers. This allows access to a broad embedded software ecosystem and further security capabilities.

Video analytics manager API and connected camera Software Development Kit (SDK)
A feature-rich set of camera processing and analytics software capabilities including multi-stream, overlay, video HDR, temporal noise reduction, low bit rate codecs, voice activation, face detection and recognition, intelligent motion detection and object tracking. With such intelligence on the camera, customers can move from expensive, server-based video analytics in the cloud to efficient camera-based analytics. By only streaming data when qualified events happen, the camera platform is designed to reduce network bandwidth requirements and help minimise cloud storage costs.

Snapdragon 625 IP camera reference design
All of the above were brought together in a 4K connected camera hardware/software platform to help manufacturers accelerate time to commercialisation. The reference design was developed in collaboration with Thundercomm Technology who will provide worldwide distribution and technical support.

"Moving video analytics processing from the server to the camera at the point of capture opens up a class of 'conscious cameras' with tremendous benefits in terms of reduced data streams, better image recognition and immediate alerts and actions," said Raj Talluri, Senior Vice President, Product Management, Qualcomm Technologies. "With advanced connectivity, and computing technologies incorporated into the camera right at the point of capture, we push more intelligence to the edge, allowing deep learning to support highly accurate object recognition."

With its expertise in imaging, video, connectivity and compute, Qualcomm Technologies is well positioned to deliver the technologies needed in IoT, and has already brought Snapdragon connected camera solutions to products such as the LG 360 CAM and LG Action CAM LTE.


You must be logged in to comment

Write a comment

No comments




Sign up to view our publications

Sign up

Sign up to view our downloads

Sign up

embedded world 2019
26th February 2019
Germany Nuremberg
Wearable Tech Show 2019
12th March 2019
United Kingdom London
AMPER 2019
19th March 2019
Czech Republic Brno Exhibition Centre
LOPEC 2019
19th March 2019
Germany Messe Munchen
RF & Microwave 2019
20th March 2019
France Paris Expo, Porte de Versailles