With the rapid development of artificial intelligence (AI) technology, cameras have become more intelligent, and can identify people and objects, while combining IoT and sensor technology to make corresponding follow-up actions, making people's lives more convenient. This article will show you the development of smart cameras and the solutions launched by onsemi.
The unlimited potential in on-device AI for camera
Cameras are already quite mature products, which were first used simply to shoot and record various people and objects under the lens. Early cameras only had the shooting and recording function, and the subsequent image judgment and interpretation still depended on manual processing, which limited the development and application of cameras. However, with the rapid development of AI technology, the system can judge the content of the captured image and carry out subsequent processing according to the functional requirements, which makes the smart cameras system have infinite application potential.
For example, when a smart camera is used for security applications, it can judge who the photographed person is, and if there is an intruder, it can give a warning and record what is done; when a smart camera is used for asset management, it can record the product content on the current shelf, and when there is a shortage of goods, it can prompt replenishment in real time; smart cameras can also be used to detect accidental falls of the elderly, crying of children, pet behaviors, etc., which provides unlimited possibilities and opportunities for the development of smart cameras.
Smart cameras combining cloud services and Internet of Things
A smart camera is not only a camera itself. Combined with cloud services, Internet of Things and sensors, smart cameras have a wide range of features and applications development prospects. According to the report released by IDC, “the rise of vision and sensor interaction” has become one of the top ten development trends of smart home market in 2021. With the rapid improvement of AI capability of cameras, they will be widely used in more smart home devices to build a complete and natural interactive experience in vision and voice together with voice assistants.
In smart camera AI applications, the most common function is motion detection and tracking. The camera can recognize the movement of objects, rotate and shoot, and automatically take photos and send messages to the owner to capture the life fragments of babies, old people and pets. In addition, with the functions of face recognition, pet recognition, fixed area recognition, etc., you can identify whether the newcomer is a stranger or not, and you can also select a specific area in the image for detection and alarm.
In the past, in order for smart cameras to perform smart judgment functions, it usually required the AI function to be built on the local side, but the end system would be too huge. The new type of smart cameras can send the captured pictures to the cloud platform for interpretation, and then send the results back, and match with various Internet of Things and sensor devices for subsequent linkage response, which enables smart cameras to develop various systems according to the needs of users.
Smart cameras support node-to-cloud platform
Seeing the development trend of smart cameras, onsemi launched the RSL10 Smart shot Camera (SECO-RSL10-CAM-GEVB), which is a complete node-to-cloud platform and can provide smart image capture for portable and low duty-cycle Internet of Things applications (including asset monitoring and security). The platform is based on the RSL10 system-in-package (SIP) module and ARX3A0 CMOS image sensor, and has various trigger modes, including time, motion detection and environmental sensor trigger (such as temperature and humidity change). The RSL10 SIP is used as the processing hub of the camera, which can connect the captured images and sensor data to the matching RSL10 Smart Shot mobile application (available on GooglePlay™ and iOS®) through Bluetooth® Low Energy.
Using the application, developers can access a variety of features, including cloud connectivity, remote configuration of sensors (thresholds and default settings), and changing the trigger mode of image capture. The application can be connected to an Image Analytics /AI engine located in the cloud, and can use the AI cloud service to detecting and identifying objects, while the mobile application returns captured images, as well as a list of identified objects and confidence levels.
The platform uses the ARX3A0 monochrome 65° DFOV IAS module, which is a compact design form-factor form for developing compact cameras based on ARX3A0 CMOS image sensor, with 1/10" sensor and 360 fps monochrome imaging. ARX3A0 provides high sensitivity in visible light and NIR (near infrared) wavelength, and has ultra-low power consumption mode, with power consumption of 3.2 mW when active. In addition, ARX3A0 can be programmed to monitor specific zones inside its field of view, and automatically take images when the scene content changes. With SunplusIT® SPCV1100A embedded platform hardware, it offers multiple color processing functions and JPEG image compression capability.
The RSL10 SIP and ARX3A0 have impressive low-power operation capability, which is realized in hardware by a dedicated NCP6925 power management chip (PMIC) (NCP6925) and other smart power management modes. Other devices that support efficient power management include FAN54120 USB-compatible linear lithium-ion battery charger and FAN49100 ultra-low Iq buck-boost voltage regulator. Thanks to these functions, the RSL10 smart shot camera can support more than one year's battery life in standby and active modes.
Smart camera with various functions and wide applications
The RSL10 smart shot camera can support multiple image capture trigger modes, and can set the interval of shooting (interval of 1 second to 1 minute). It can shoot continuous video streams at a speed of less than 1 frame/second on compatible mobile devices supporting BLE5.0 and higher, and supports proximity detection, acceleration trigger and environment change trigger. Through the Bluetooth Low Energy connection (Bluetooth 5) enabled by the RSL10 SIP, firmware Over The Air (FOTA) can be performed. It is available in small form factor (55 x 6 mm).
The RSL10 Smart Shot Camera enables automatic exposure during image processing, supports the RSL10 Smart Shot mobile application (GooglePlay™, iOS®), and is connected to the Amazon Web Services™ (Amazon Rekognition) cloud to detect and identify objects in images through Image Analytics /AI engine. This portable, node-to-cloud and near turnkey design with long battery life (>1 year) can shorten the product development time, is suitable for multiple trigger functions of various use cases, and can be applied to asset monitoring, security, smart agriculture and criteria-based monitoring.
Typical applications of the RSL10 smart shot camera include adding smart cameras to wearable safety equipment, such as hard hats. In a commercial environment, smart cameras can be used to monitor the contents of shopping carts to support automatic checkout. In vehicles, smart cameras will be used to monitor the occupants, such as children in rear seats, to provide early warning of any safety issues. At home, smart cameras can scan items in the cupboard to create a shopping list, which are quite widely used.
Conclusion
Combining cloud platform, Internet of Things and sensors, the smart cameras have brought unlimited development potential for various applications. The RSL10 smart shot camera launched by onsemi is a complete solution and development platform, which can be adjusted and expanded according to application requirements, has a wide range of application development space, and is the best solution for developing smart home and related applications.