Hikvision Wins CIIF 2026 Industrial Automation Award for AI Ultrasonic Inspection, Unveils Industrial Digitalization Products

Leiphone.com reported on Oct. 15 that Hikvision’s AI ultrasonic inspection system uses an ultrasonic large model to model raw ultrasonic signals, perform quantitative analysis and classify defects. The approach turns ultrasonic testing, traditionally dependent on certified specialists and manual waveform reading, into a standardized and automated process. The technology is intended to lower the threshold for ultrasonic inspection, which has mainly been used in special industries such as aerospace, and to bring it to new energy vehicle, semiconductor and other industrial component testing.

To enable AI to read product quality, Hikvision addressed three difficulties, according to the report: strong clutter in signals that can obscure weak defects and raise missed detection rates; poor algorithm generalization when materials change; and difficulty in identifying and quantifying defects on complex workpieces. The company developed a knowledge-enhanced signal reconstruction network and ultrasonic large model with four functions: adaptive noise separation to make weak defect signals distinguishable; embedding experienced inspectors’ waveform-reading know-how into algorithms; self-supervised pretraining and few-shot transfer for cross-material and cross-workpiece generalization; and automatic quantitative analysis that calculates defect location, size, type, area and weld depth. Phased-array recognition accuracy reaches 0.1 mm, and the system can identify 5-micron-level defects in MLCC and other micro components, the report said.

Hikvision’s ultrasound product manager said the AI ultrasonic product can independently complete full-process interpretation and achieve 100% automatic interpretation in typical scenarios. In new energy vehicle component inspection, AI deep learning algorithms enable end-to-end automatic determination of defect type, position and size, improving single-workpiece inspection efficiency by more than 60% and complex workpiece inspection efficiency by more than 100%, the report said. This allows ultrasonic nondestructive testing to move from sampling to batch full-inspection tempo and reduces manual image review and repeated inspection costs. In one actual application project, defect recognition accuracy for new energy vehicle components was at least 99.5%, with a missed detection rate of no more than 0.5%.

In a deployment model combining AI analysis boards with customized equipment integration, overall enterprise investment is reduced by more than 50% compared with directly purchasing complete machines, according to the report. By reducing destructive metallographic sampling, sample and material loss falls by 80%, and costs related to rework, scrap and quality handling are further reduced. Hikvision has formed a product series including ultrasonic microscopes, three-axis, four-axis and six-axis automated equipment, intelligent ultrasonic all-in-one machines, portable devices and multiple probe types. Applications include automotive friction stir welding, MLCCs and wind turbine composite blades.

At CIIF, Hikvision displayed products for quality inspection, equipment monitoring, production operations and safety production. Its production-line behavior SOP inspection solution comes with ready-to-use algorithms for common production lines, provides real-time operation visibility, alerts on abnormalities and guides workers to correct them, the report said. For new workpieces and scenarios, only one to eight images are needed to register and generate a new model for minute-level deployment. With zero-code orchestration software and an LA client, the company said users can build AI inspection solutions to handle complex production-line control needs.

Hikvision Rayin launched a high-precision 3D AXI inspection equipment for semiconductor packaging. The equipment performs 3D image reconstruction and defect analysis and is equipped with an X-ray industrial nondestructive testing large model. It can automatically identify micro defects inside chips that are as small as one-hundredth of a hair, with detail resolution below 1 micron and geometric magnification up to 2,500 times, bringing inspection capability to the sub-micron level. Wang Wei, president of Hikvision Rayin, said the product can also enter research institutes and laboratories as metrology equipment, meaning the company’s products are extending toward higher-precision measurement. The report said Hikvision’s ultrasonic nondestructive testing series can inspect inside chips with ultrasonic microscopes for micron-level defects.

In equipment monitoring, Hikvision introduced a positioning acoustic fiber sensor with an industrial acoustic diagnosis large model to monitor belt idlers around the clock, automatically identify idler looseness and bearing damage, and locate anomalies precisely. In high-temperature production, Hikvision Microimaging’s four-in-one high-definition visual pyrometer, equipped with the Hikvision Guanlan industrial temperature measurement large model, combines high-precision temperature measurement and high-definition video. It addresses overexposure and color cast under strong high-temperature light and can clearly restore details at temperatures up to 3,200°C, according to the report.

Other exhibits included infrared temperature measurement analysis cameras for real-time temperature measurement of moving workpieces; a multimodal large-model Cloud Eye agent that can perform factory patrols and energy reviews through voice commands; six new barcode PDA models with AI large models that can read DPM codes on metal parts in CNC workshops with cutting fluid splashing and barcodes on 10-meter-high racks from the ground; a ChatBI data analysis agent that takes tasks, conducts analysis and produces reports; a TDLAS gas cloud imaging remote sensor that integrates laser monitoring, gas cloud imaging and visible light with a large-model enhancement algorithm to improve micro-leak recognition; safety production large model algorithm software with more than 70 algorithms available out of the box and open interfaces for existing video platforms; and the SafetyStar monitoring and warning client, which offers a cost-effective standalone deployment for small and medium projects. Front-end cameras with large models can identify hazards such as unauthorized entry into dangerous areas, failure to wear required protective equipment and lingering in robotic arm work areas, and intervene immediately.

Hikvision said it began a strategic adjustment a decade ago, extending its technology system from visible light to multi-dimensional sensing that covers millimeter wave, infrared, X-ray, ultraviolet, acoustic and other physical sensing technologies. Its product line has also expanded from security to non-security applications, with products focused on quality inspection, equipment monitoring, production operations and safety production. According to the report, cumulative shipments of Hikvision Robotics’ machine vision products have exceeded 10 million units, and mobile robots have rolled off the line more than 200,000 units. One in every two industrial cameras and one in every three mobile robots in the domestic market comes from Hikvision Robotics, the company said. It has served more than 20,000 customers and deployed in more than 1,000 segmented application scenarios, covering automotive parts, 3C electronics, lithium batteries, pharmaceuticals, tobacco, footwear and apparel, and photovoltaic industries.

Hikvision Rayin entered industrial X-ray inspection in 2021. Its business now covers electronic manufacturing, food and other sectors, and it ranks among the industry’s first tier and among China’s top three in several segments, according to the report. Using AI, it has tackled difficult inspection scenarios, including X-ray plus AI inspection of folding-screen phone hinges, upgrading manual magnifier checks to automated full inspection; after adoption by a leading phone maker, the solution expanded across the mobile phone industry. Hikvision Microimaging’s four-in-one visual pyrometer has been widely used in semiconductor, industrial materials, steel and nonferrous metals production, and extended to automotive casting and university research, the report said. Hikvision also said it applies its own technologies and products first in its manufacturing bases. Its SOP inspection solution has been deployed at more than 600 core workstations in its Chongqing, Tonglu and Wuhan production bases, improving efficiency at a single workstation by up to 80% compared with manual methods. In 2025, Hikvision’s smart manufacturing factory was selected for China’s first batch of leading-level smart factory cultivation list. The SOP solution has also been widely deployed in Li Auto’s supply chain. At a seat supplier in Changzhou, the seat C-ring missed-nailing rate fell to as low as 0.01%, and the supplier reused the solution with its upstream and downstream suppliers. At Lianhe Dongli, a powertrain supplier, the solution covered 100% of key work processes for new energy vehicle electric drive power quality, the report said.

“Industrial digitalization and intelligent upgrading begin with every real scenario,” a Hikvision official said, according to the report. “From production process coordination to a production line, a piece of equipment or a work site, Hikvision is committed to deeply integrating IoT sensing and AI into thousands of industrial scenarios to continuously help industrial scenario digitalization and create more value for users.” The official also said digitalization is the foundation of industrial intelligence and remains an ongoing process.

Editor’s Summary: Hikvision used CIIF 2026 to show how AI and IoT sensing are being applied to industrial inspection, monitoring, production operations and safety. Its AI ultrasonic inspection won the Industrial Automation Award, with reported accuracy of at least 99.5% and missed detection of no more than 0.5% in new energy vehicle component testing, while new X-ray, acoustic, temperature measurement and safety products extended large-model applications into specific factory workflows. The company also reported broad deployment of its SOP inspection system in its own bases and in automotive supply chains.