The advent of Industry 4.0, coupled with the rapid advancement of AI, big data, the Internet of Things, robotics and other related technologies, is driving the comprehensive intelligence transformation of advanced manufacturing. Alongside this trend, technological products are imposing increasingly stringent requirements on advanced manufacturing: more process control points, faster production speeds, and growing complexity in the types and parameters to be inspected. Inherent production waste cannot be fully eliminated, while manual quality inspection, hard to replace, carries risks of missed defects. Product changeovers on production lines involve a rising number of workstations and equipment, posing greater challenges for flexible production management.

Faced with three major industry challenges of "quality improvement, cost reduction and enhanced flexibility", advanced manufacturing is in urgent need of critical support from inspection technologies and yield management. This helps fully unlock the potential of production capacity and yield, and guarantees safe and reliable production.

Leveraging its full‑stack self‑developed core technology platform covering optics, mechanics, electronics, computing and software, together with industry‑leading process inspection and yield management solutions, Jingce Electronics proudly launches the AIVIS Industrial‑Grade Intelligent Vision Inspection Workstation. Deeply integrating Convolutional Neural Networks (CNN), Transformer architecture, as well as innovative self‑developed zero‑shot anomaly detection, defect generation, self‑classification technologies and few‑shot learning algorithms, the workstation embeds five collaborative subsystems to cover the full workflow of defect detection, data annotation, model management and iterative learning. It delivers high‑performance end‑to‑end intelligence spanning optical imaging, image processing, defect detection, data analysis, model iteration and process optimization. Providing risk‑free and optimized intelligent production solutions for industrial vision inspection, it features outstanding stability, strong generalization capability and high cost‑performance, and has been widely deployed in quality‑inspection scenarios for display, new energy, semiconductor and 3C industries.

Distributed Plug‑in Management for Flexible Deployment

Adopting a modular architecture, the AIVIS Industrial‑Grade Intelligent Vision Inspection Workstation breaks down the complex system into relatively independent, manageable and flexibly‑invokable modules to enable flexible assembly of functional plug‑ins. Its five subsystems — intelligent inspection platform, algorithm development platform, data annotation & model training platform, application platform and full‑scenario AI toolchain — are deployed in a distributed manner. The training platform supports one‑to‑multiple inspection deployment, reducing redundancy and customized R&D workload, and better adapting to production management of diverse products on the same production line.

Intelligent Inspection Platform: Technological Innovations Enable Efficient Defect Detection to Boost Yield

Integrating data acquisition, image pre‑processing, algorithm‑driven inspection and automatic reporting, the AIVIS Intelligent Inspection Platform applies unsupervised anomaly detection and self‑supervised classification detection technologies. It delivers risk‑free and optimized intelligent production solutions for industrial vision inspection, helping manufacturers efficiently identify and handle defects and drive yield improvement.

Algorithm Open Platform: Open and Extensible Architecture for WYSIWYG Algorithm Development

The AIVIS Algorithm Development Platform features an open operator‑extensible architecture that flexibly supports operator development, integration and expansion. Equipped with a visual algorithm editor, it breaks the traditional "black‑box mode" of algorithms. Users can build algorithm execution logic flowcharts on the algorithm canvas by dragging‑and‑dropping operators without programming skills, realizing true WYSIWYG (What‑You‑See‑Is‑What‑You‑Get) algorithm development. It supports modular algorithm extension and deployment to enable algorithm implementation across all scenarios of industrial vision inspection.

Data Annotation and Model Training Platform: Optimal Parameter Configuration to Boost Algorithm Development Efficiency

Application Platform: Real‑time Monitoring & Alerting to Ensure Stable Production on Production Lines

The AIVIS Application Platform leverages AI‑based visual recognition to deliver feedback‑enabled, storable and replicable standardized and streamlined inspection. Through intelligent monitoring, data analysis and anomaly surveillance, it supports fault alerts, equipment anomaly prediction, process yield early‑warning and emergency response mechanisms, ensuring stable production of manufacturing lines.

Full‑Scenario AI Toolchain: Realize Intelligent Closed‑Loop from Data to Model

The AIVIS Full‑Scenario AI Toolchain delivers an end‑to‑end solution spanning from data management to model deployment. It integrates intelligent defect generation, anomalous data acquisition, data cleansing, data quality assessment, model evaluation and automatic model iteration capabilities. It greatly improves the efficiency for development, training and deployment of industrial quality‑inspection models and applications, enabling traceable quality and self‑iterating models.

High Uptime, Dramatically Reduced Model Changeover Time, Industry‑Leading Defect Detection Capability

Leveraging zero‑shot learning, self‑classification and automatic model iteration technologies, the AIVIS Industrial‑Grade Intelligent Vision Inspection Workstation supports rapid go‑live with a baseline detection rate of 80%.

It adopts coordinated operation of both unsupervised and supervised technologies for distribution, deployment, upgrade and management of quality‑inspection capabilities on customer‑side devices. Powered by unsupervised technologies, it achieves 70% defect detection and cuts model changeover time by over 80%, while maintaining stable performance in complex scenarios. With nested supervised technologies and data checks, it accomplishes high‑precision identification and classification for the remaining complex defects, delivering industry‑leading defect detection performance.

Guided Inspection Solution & Zero‑basis Operation for Convenient Maintenance and Trial Use

Addressing industry pain points of conventional industrial vision inspection such as complicated operations, high technical barriers and human‑induced errors, the AIVIS Industrial‑Grade Intelligent Vision Inspection Workstation delivers guided configuration.

With standardized workflows, human‑centered interaction and accumulated knowledge base, it adopts step‑by‑step guidance to help users effortlessly complete product model configuration. It lowers operating costs and enables convenient task execution. Particularly suitable for scenarios requiring multi‑variety production, high precision and robust traceability, it can be rapidly adapted to diverse industrial products and production environments without extensive re‑development work, featuring excellent usability and easy maintenance.

Backed by outstanding technical strength and scenario‑adaptation capabilities, Jingce Electronics AIVIS Industrial‑Grade Intelligent Vision Inspection Workstation fully meets customized scenario requirements for industry customers. It has been deployed in over 100 smart factories, with more than 1,000 units of the AIVIS Standard Edition rolled out in volume. It serves high‑value sub‑scenarios in the semiconductor, display and new‑energy industries, including crack inspection, appearance inspection, pattern quality inspection, and AI re‑judgment.

| Screen Pattern Quality Inspection: Built on the technical architecture of "base large model + few‑shot inspection", AIVIS breaks through the bottlenecks of traditional screen inspection to capture more defects, achieving a detection rate of 99% and completing few‑shot iteration within one day.

| Module Appearance Inspection: Leveraging high‑resolution industrial cameras and deep‑learning algorithms, AIVIS unifies appearance‑inspection data across diverse manufacturing processes to realize quality traceability and efficiency improvement, meeting the requirements for "zero‑defect" manufacturing.

| New‑Energy Lithium‑ion & Hydrogen‑powered Material and Finished‑Battery Inspection: Adopting the solution of "multi‑model collaborative architecture + attention‑mechanism guidance + few‑shot reinforcement learning", AIVIS delivers outstanding zero‑miss‑detection performance, with the over‑inspection rate precisely controlled at an ultra‑low level of 0.05%.

| Semiconductor Wafer Inspection: By adopting "extreme model pruning + attention‑mechanism guidance", AIVIS substantially reduces the miss‑detection rate. It achieves over‑inspection <0.2% and miss‑detection <0.002% (20 PPM), reaching an overall advanced industry level.

For a long time, Jingce Electronics has jointly built a Joint AI Laboratory with Huazhong University of Science and Technology to continuously advance the research project of “Fundamental Platform for AI‑based Complex Visual Defect Inspection”. Both parties have co‑developed the world’s first zero‑shot industrial defect‑inspection method requiring neither training nor prompts. It innovatively employs self‑supervised training to learn defect representations and parametric classifiers. Accurate defect‑category identification can be realized without manual annotation, enabling autonomous high‑precision classification of industrial defects.

Taking independent innovation as the core and industry‑university‑research collaboration as two driving pillars, Jingce Electronics leverages the Joint AI Laboratory to build an intelligent inspection solution centered on the AIVIS Industrial‑Grade Intelligent Vision Inspection Workstation and supported by a full portfolio of hardware and software products. The solution addresses customized inspection requirements across diverse industries and production lines, enables seamless connection between inspection data and process data, and fully empowers production‑line process management and yield management. It helps enterprise customers achieve core goals of quality improvement, cost reduction and flexibility enhancement, and brings about paradigm innovation for intelligent manufacturing.

As a pioneer in industrial‑grade vision‑inspection technologies, Jingce Electronics takes the lead in proposing to realize “modularized manufacturing processes” and “intelligent defect inspection” through “digitalized quality standards” and “digitalized inspection modes”, and ultimately build a “yield‑management system” that underpins advanced manufacturing. Serving as a key enabler of Jingce Electronics’ yield‑management solution, the AIVIS Industrial‑Grade Intelligent Vision Inspection Workstation delivers yield‑management capabilities across multiple industries and fields. It supports Jingce Electronics in growing into customers’ preferred comprehensive‑solution provider focusing on test equipment for the semiconductor, display and new‑energy industries.