Powerful 2D+3D+AI Algorithm Library

With modular design, the algorithm library is built up with loosely coupled classes and UI components. Customized logic can be achieved by simple combination of these elements. Error rate and exceptions can be well controlled because these elements were stabilized with repeated tests and upgrades.

• 2D Algorithm Library
  This library includes image acquisition, image pre-processing, edge extraction, image matching, area analysis, geometry, calibration and related modules. All algorithms support re-entry and hardware acceleration.
• 3D Algorithm Library
  This library includes data acquisition, 3D transform, point cloud fusion, resampling, matching, region fitting, 3D geometry and related modules. Most of them support hardware acceleration.
• Deep Learning Algorithm Library
  This library focuses on classification, detection and segmentation area which includs training and inferring of AlexNet, ResNet, SSD, YOLO, FCOS, UNet, UNet++, SOLO, etc. Various labeling tools are provided, including points, lines, polygons and sketching. Labeling results can be exported to JSON, pictures and TXT. Both local training and cloud training are supported. And both inference at the edge with GPU or CPU and in the cloud are supported.

Flexible and Extensible Basic Program Framework

InteVega applies cutting-edge software framework models and maintains a low degree of coupling among modules. Since the function expansion does not affect the functions of the previous version, functions of various business modules are continuously improved and expanded independently without much concern for the increase of regression errors. Each module contains built-in independent unit tests and performance tests to ensure the stable and efficient operation of basic function.

Integrated Scenario-Based Application Solutions

The data and empirical parameters amassed from projects are integrated into the platform during version update, such as defect detection under complex backgrounds, and common algorithms and models abstracted in panel detection scenarios, new energy application scenarios, semiconductor detection scenarios and other applications can be quickly deployed to new projects. Applications such as multi camera fusion, camera laser data fusion, online precision measurement and so on can be quickly applied and stably operated on the InteVega platform.

Friendly Interactive Logic Design

With the data flow architecture, InteVega encapsulates the algorithm into modularity, and visualize the operation so as to complete the configuration of projects only by dragging and connecting without programming. In Addition, for specific application scenarios, it integrates the algorithm package, which can be deployed with one click, making it easy to use. It supports 4-level programming of operators, application engineers, module engineers and developers. Different levels access different authority, which greatly facilitates the routine application settings.

Excellent Testing and Diagnostic Tools

The InteVega platform internally realizes data recording and tracking mechanism, which can fully record and feedback the details and problem of long-time operation on site, helping on-site engineers quickly locate and solve problems. No matter who designs the business logic, the algorithm platform ensures that the algorithm problem is 100% reproduced and solved by remote expert consultation, ensuring the long-term and low failure rate operation of the equipment.

Substantial Data Visualization and Output

InteVega supports a variety of graphs such as curves, histograms, pie charts, etc., which facilitates the graphic display of data, the output and reading of XML, INI, TXT, JSON and other file formats as well as the writing and visual display of data and picture logs. In addition, this product also supports the output and reading of XML, INI, TXT, JSON and other file formats, as well as the writing and visual display of data and picture logs.