Data processing includes classification and labeling. Image classification supports single-label and multi-label. Image annotation include the methods of points, rectangles, polylines, polygons, line segments and circles, and support various types of annotation tasks. It also supports pre-created annotation templates for multiple points, annotates key points in the target object through a fixed number of key point groups, and the connection relationship between points, automatic average distribution, point attributes, etc. Further improve annotation efficiency with eyedropper and format brush tools.

Select the algorithm type and train the model with the uploaded data. After the model training is completed, the model effect can be verified online. It supports the selection of a variety of algorithms to meet the different needs of performance and effects in different scenarios. It also supports importing your own algorithms, small target detection and other precision optimization functions. Based on our unique pre-training model, a high-precision model with excellent performance can be obtained by training with a small amount of data.

After the training is completed, the model can be deployed on a public cloud server or a private server, or packaged into a device-side SDK that can run offline, or purchased a software-hardware integrated solution directly. It can be flexibly adapted to various scenarios and operating environments, and can also be directly released as a device-cloud collaborative deployment package and delivered to edge devices for application.