Edge Computing Box as an Application Product of RK3588 - for Open Kitchen Monitoring, Construction Sites, Industrial Production, and OCR

By equipping the edge computing box with multiple cameras and AI algorithms, different local AI visual recognition projects can be developed, including the popular Open Kitchen Monitoring platform in China, smart construction sites, industrial production, OCR license plate recognition, and other common project developments.

Dingchang provides API platforms and DEMOs with a variety of different algorithms, which can help cooperative customers within the ecosystem quickly get familiar with the AI algorithm development process of the RK platform. It offers up to 35 visual algorithm DEMOs that can be learned and used commercially. Alternatively, customers can directly use the data interfaces of the application layer API provided by Dingchang to obtain algorithm results, eliminating the need for them to carry out underlying development such as algorithm transplantation and training.

1. For example, in the development of the Open Kitchen Monitoring platform.

It can be combined with Dingchang's RK3588 edge computing box, other third-party brand POE cameras, and Dingchang's AI algorithm platform or API. Or use the deep learning algorithm ResNet provided by us to identify the ingredients in the kitchen, conduct various detections on tableware and the kitchen environment, and monitor some violations of kitchen staff, etc. Or use the YOLO algorithm provided by us to detect the cleanliness of the kitchen, foreign objects, clutter, cockroaches, etc., and YOLO is also suitable for monitoring the behaviors of kitchen staff, such as whether they are wearing work clothes, whether they are operating according to the procedures, etc.

2. As for the algorithm development for smart construction sites.

The application scenarios of construction site algorithms are more complex, and different construction site scenarios also vary. It includes dozens of recognition scenarios such as common safety helmet recognition, human shape detection, flame detection, target tracking, safety visual analysis, personnel stay detection, fall recognition, and reflective work clothes recognition. Dingchang's YOLOV8 series of algorithms can be used for the above-mentioned recognition development.

3. In the algorithm development for industrial production.

Such as real-time production data analysis of the automatic food production line, product quality defect detection, whether there are foreign objects in food packaging, detection of defects in food raw materials or finished products, and whether the production line processes are normal, such as detecting whether the cans are in the correct position, etc. Algorithms like YOLO V5, YOLOV5_seg, YOLO V8, YOLO V8_seg, etc. can be used for development.

4. For OCR license plate recognition.

It is recommended to use LPRNet for the application development of the license plate recognition algorithm. Cameras such as USB, MIPI, or POE cameras can be selected for the camera configuration. Generally, license plate recognition consists of four processes: license plate positioning, character segmentation, license plate recognition, and license plate display.

The above is just a small part of the functions of Dingchang's RK3588 edge computing box summarized by our editor.

For more details, you are welcome to call or add our customer service for further information. We are dedicated to serving you.