Run Object Detection using Deep Learning on Raspberry Pi 3 (2)
BerryNet: Turn Raspberry Pi 3 Into Intelligent Gateway
In the previous post, we talked about The difficulty to use deep learning for object detection on low-power devices such as Raspberry Pi. Before going deeper into more numbers and analysis, let’s do something practical; to use Raspberry as the intelligent gateway of cameras and IoT devices. With the intelligent gateway, you can connect IP camera to Pi to monitor your kid and get notifications when someone is nearby.
DT42 released the open-source project, BerryNet [1] to turn Raspberry Pi 3 into an intelligent gateway with deep learning running on it. No internet connection is required, the analysis is done locally on the Raspberry Pi 3 itself.
Figure 1 shows the software architecture of the project, we use Node.js, MQTT and an AI engine to analyze images or video frames with deep learning. So far, there are two supported AI engines, the classification engine and the object detection engine. The current supported models are listed in Table 1.
One of the applications of this intelligent gateway is to use it for the security purpose. For example, Figure 2 shows the analyzed results from the camera hosted in the DT42 office. The frames were captured by the IP camera and they were submitted to the AI engine. The output from the AI engine will be shown on the dashboard.
By adding more AI engines to the intelligent gateway, it can analyze not only videos but also voices or data from other sensors. We are also working on adding IM and Email notification to the BerryNet project so you can get a notification when there is a dog showing up in the office meeting area (although, I guess a normal office probably won’t need this function).
With BerryNet, IoT will not only stand for Internet of things but also Intelligence of Things.
At DT42, we believe that bringing deep learning to edge devices is the key towards the future. It not only saves costs of data transmission and storage but also makes devices able to respond according to the events shown in the images or videos without connecting to the cloud. In the future, BerryNet will be a place to collect information, a center to analyze the signals locally and a smart brain for IoT devices.
The project is licensed with GPL because we also believe that deep learning and AI will be democratized for everyone [4] and every device.
[1] https://github.com/DT42/BerryNet
[2] https://arxiv.org/pdf/1512.00567.pdf
[3] https://pjreddie.com/media/files/papers/YOLO9000.pdf
[4] https://medium.com/@matelabs_ai/why-do-we-need-the-democratization-of-machine-learning-80104e43c76f
