Artificial intelligence AI

      ADR is a self-developed intelligent defect rejudgment system based on AI+ traditional vision algorithm, which is used to collect the traditional inspection images of field equipment and intelligently classify them through neural network algorithms, rejudge the existing detection results, increase the detection scope and indicators, reduce labor, unify quality control standards, and improve work efficiency. The overall deployment and management of the factory can provide real-time warning of production process quality and improve production management process and production quality.

设备终端图像获取Device terminal image acquisition

AI模型推理分类AI model inference classification

服务器数据对接Server data interconnection

跨终端预警控制Cross-terminal early warning control

GPU-AI服务器GPU-AI server

AI复判系统AI rejudgment system

光学图像检测软件Optical image inspection software

AI训练软件AI training software

Product features:
1.Efficient defect re-judgment
obtains NG images of traditional AOI inspection equipment across terminals, classifies them through AI model inference, and intelligently determines the results of existing inspection items.
2.Digital equipment early warning through AI intelligent analysis, real-time feedback data to the customer system terminal, digital production data statistical presentation, for traditional equipment production early warning.
3.Standardized digital interface

The standardized interface can realize the deployment of the system under the premise of making minimal changes to the production line and equipment, quickly connect with the customer terminal, and save deployment time and cost.
4.Rapid deployment capability can be flexibly deployed in the production line, rapid migration capability
with the line, batch replication of the production line, and rapid upgrade.
5.Modular system

Therejudgment system is modular, and can be quickly docked according to the different needs of customers, and can be flexibly deployed with different terminals such as equipment or system servers.
6.The efficient model algorithm
only needs a very small number of images to train an efficient model, shortening the deployment cycle and training difficulty.
Applications:

半导体semiconductor

面板显示Panel display

新能源New energy

汽车Car

食品foodstuff

医药medicine