Cognex, a maker of machine vision systems for industrial automation, has seen its shares surge 85% this year as AI extends into physical applications. The company produces cameras and software that enable robots to detect, measure, and position objects, combining vision technology with AI to enhance automation. In the second quarter, Cognex reported earnings of 45 cents per share on sales of $291 million, with earnings up 80% and revenue up 17% year-on-year. Both figures exceeded analyst expectations. The company recently partnered with Nvidia and Qualcomm to launch AI vision platforms. Analysts have raised full-year profit estimates to $1.68 per share, representing 65% growth. Mutual funds collectively own 67% of outstanding shares, with increasing holdings over recent quarters.

AI-Powered vision systems transform plastics quality control. Artificial intelligence is enabling vision systems to detect subtle plastics defects in real time, reducing scrap and improving quality control. August 17, 2026 Companies like Krevera and Cognex are deploying advanced AI vision systems that reduce scrap rates and improve manufacturing efficiency. AI's ability to analyze data at lightning speed is proving especially useful in vision systems used to detect part defects in real time in plastics manufacturing. Traditional vision systems have relied on cameras placed in key locations in automated workflows to detect part presence and orientation. With the addition of AI, the real-time detection of part defects is improving exponentially, offering manufacturers unprecedented quality control capabilities. For instance, Krevera has created a proprietary vision system for plastics injection molding that is trained on "a larger data set than any of our competitors," explained cofounder and Chief Content Officer Sebastian Schneeweiss. Where visual detection gets complicated, Schneeweiss continued, is flagging what he calls chemical defects. A typical vision system will check that "your six pack of beer has six beers inside of it. Super easy. But if you're checking that your bottle cap has a malformation, while you're looking at all six, it gets much more difficult." Cognex, too, is creating from-the-ground-up AI-driven vision systems that let manufacturers "detect cosmetic and structural defects earlier, reduce scrap and rework, improve yield, limit false rejects, and perform 100% inline inspection at production speeds," said Georges Gauthier, senior manager, product management. Plastics manufacturing is particularly suited to AI vision inspection, Gauthier continued, "because many common defects are inherently variable. Molded parts may be reflective, textured, colored, or transparent, making them difficult for traditional rule-based systems to inspect consistently. AI is effective in these environments because it can learn from examples and adapt to real-world variation." Krevera gets granular with defect detection. To flag chemical formations, Krevera's AI-driven vision system steps beyond non-AI vision systems that need constant calibration. Krevera has trained its system "on all sorts of dust or dirt on the lens, almost any type of lighting configuration you can imagine, and every type of possible short shot or flash defect on a product," Schneeweiss confirmed. Krevera's unique system can detect a variety of flash defects, including: - Excess plastic - Short shots - Splay caused by water in the plastic or a screw shear issue - Burn marks when molds can't vent properly, so pressure causes chemicals to combust The complex circumstances causing flash defects "have very random ways of coming out of the mold - they're going to look really weird. One short shot can look completely different than a million other short shots." For a traditional vision system to detect such fine defects, it must be ultra-sensitive and recalibrated about every two weeks, Schneeweiss continued. Even then, dirty lenses interfere with accurate detection, flagging defects that aren't there. "Let's say you're working in a dirty factory and there's dust or dirt on the lens," Schneeweiss said. "The vision system's detecting everything as defective. If there are lighting changes, all the pixels are technically slightly different, and the system is going to detect everything as defective. Typically, having really high sensitivity in a vision system usually means you're going to have a bunch of false positives." Manufacturers will attempt to overcome that issue by putting cameras exceptionally close to where defects are likely to occur. "If you get the camera right up to where the defect could form and the defect forms, it's really easy to detect because 30% of the pixels are off. But if you're looking at a 27-gallon tote or a 5-gallon bucket, which are the typical products we look at, and you have a defect the size of a square millimeter but your camera's 3 feet away, the system's looking for a defect in less than 1% of the screen real estate in a defect." Amid production issues like dust and dirt on the lens, lighting changes, product orientation changes, and camera shifts, all those issues change the image significantly more than 1% - "if there's a 1% change in this entire image, I have to say it's defective." Krevera's AI-based vision systems are already paying big real-world dividends for injection molders. "We focus on three problems: scrap, labor, and machine uptime. Those are really the three killers to our clients. For our typical client, we can reduce their scrap by about 70% to 90%. We can reduce their direct labor by roughly 80% because of automation improvements. These numbers come from directly from the finance teams of our clients." AI-first systems from Cognex deliver hybrid approach. Another company taking an AI-first approach to its vision systems and training them on real production images is Cognex, whose newest vision systems include the In-Sight 3900. "Plastic manufacturers face a unique inspection challenge because many defects are easy for a trained operator to recognize but difficult to define with fixed rules," Gauthier said. "Scratches, sink marks, flash, short shots, discoloration, burn marks, and subtle surface blemishes can vary by material, color, texture, transparency, and finish. Cognex addresses this challenge by using AI-powered inspection tools that learn from real production images, helping manufacturers distinguish acceptable variation from true defects. Because AI processing runs directly at the edge, these inspections can be performed at production speeds without sacrificing accuracy or throughput." Featuring AI designed "as a core capability, not as an afterthought," Cognex combines advanced AI tools with traditional rule-based machine vision in a single system, giving manufacturers the flexibility to use the right inspection method for each task. "For plastics manufacturers, that means one inspection cell can use conventional vision tools to verify dimensions or assembly features while using AI to identify cosmetic or variable defects that are difficult to capture with rule-based programming," Gauthier explained. "Rather than replacing traditional machine vision, AI expands the range of plastics applications that can be automated reliably. This hybrid approach is especially valuable in production environments where product appearance varies significantly from part to part." One example: A consumer packaging manufacturer producing injection-molded caps and closures can use AI-powered vision to inspect every part in real time. The system can distinguish normal manufacturing variation from true defects, helping maintain quality standards while reducing reliance on manual inspection. The future of AI vision in plastics manufacturing. As AI progresses in tandem with Industry 4.0 and 5.0, "we expect adoption in plastics manufacturing to accelerate over the next several years as manufacturers face growing pressure to improve quality, reduce waste, address labor constraints, and support more flexible production," Gauthier noted. "We see AI vision becoming a foundational technology for Industry 4.0 and Industry 5.0 initiatives. Before a robot can make a decision or an analytics platform can optimize production, manufacturers need reliable information about product quality. AI vision provides that information in real time by identifying defects, generating inspection data, and helping manufacturers make faster, more informed decisions. In plastics manufacturing, that means higher yields, less waste, and greater confidence that every product leaving the line meets quality standards." In fact, the most recent Cognex release is targeted directly to the connected factory floor and features dual communication ports enabling real-time factory floor decision-making and data exchange with higher-level enterprise monitoring systems. Geoff Giordano is a tech journalist with more than 30 years' experience in all facets of publishing. He has reported extensively on the gamut of plastics manufacturing technologies and issues, including 3D printing materials and methods; injection, blow, micro and rotomolding; additives, colorants and nanomodifiers; blown and cast films; packaging; thermoforming; tooling; ancillary equipment; and the circular economy. Contact him at [email protected]. Want more PlasticsToday in your search results? Editor's choice. Aug 18, 2026 Aug 17, 2026