Qualcomm releases an IP Camera Development Platform with Machine Vision Capability

As Qualcomm continues to support developers, it recently announced its first IP Camera development platform to enable developers to leverage Qualcomm’s Snapdragon Processors in new ways.

Known as the Snapdragon 650 IP Camera this is a reference design and development platform for much more than simply recording or streaming video, it enables developers to quickly get up and running with a powerful platform for video analytics, machine vision and other compute intensive tasks.

By integrating 4K HEVC encode and decode capability, comprehensive connectivity, and advanced analytics within the camera.

Currently most IP cameras are used for surveillance or recording events.

What sets Qualcomm’s new IP Camera apart from the rest is that it contains a Snapdragon 650 Processor, providing not just 4K quality imaging acquisition but powerful processing as well.

Featuring a 64-bit hexa-core CPU (running at 1.4GHz), Qualcomm’s Adreno GPU (Graphics Processing Unit), and the Qualcomm Hexagon DSP (Digital Signal Processor), The Snapdragon 650 IP camera can be used to “understand” what its looking at such as traffic patterns, Employee Identification for security and conditional access, all with a high level of reliability and accuracy. All of this on-board capability will enable developers to rapidly employ new technology trends in Artificial Intelligence, Deep Learning and machine vision.

Example of face detection — courtesy Google Images

By combining this computing power with a high quality, dual image sensor, featuring a 21 Megapixel camera from Sony, 4K HVEC (H.264/H/265) compression can be performed within the IP camera module, reducing bandwidth consumption by 50 percent.

For connectivity, the Snapdragon 650 IP Camera includes GPS (for location awareness), and 802.11a/b/n/ac Wi-Fi, Ethernet (with PoE), Bluetooth Smart 4.1, and LTE (cellular) providing a fully mobile capability.

Running the Android OS 5 (Lollipop) and using the Android SDK/NDK, developers can quickly implement functions such as motion detection, face detection, facial recognition, and other video analytics. With 2GB of RAM and 8GB of Flash Memory, there is plenty of storage for application code and images.

The IP Camera Development Platform is available now from Thundersoft, and there’s a slightly different configuration called the Open-Q™ 650 IP Camera Reference Design available from Intrinsysic.

Image courtesy of ThunderSoft
Open-Q™ 650 IP Camera Reference Design from Intrinsyc

Be sure to check out The Qualcomm Developer Network (QDN) Forums, along with 96Boards for additional information as it becomes available, and the QDN Downloads section for specific SDK’s for Adreno, Hexagon, and Computer Vision.