Transforming Defect Detection with Vision AI

When it comes to managing your bottom line, where do you start? Lighten up on the team lunch, take a fine-tooth comb to marketing spend, audit the necessity of overtime hours. Anything to shave a few dollars from the overhead. With ongoing economic uncertainty, especially in the manufacturing sector, it’s timely to take a closer look at the expense sheet. An underrated place to look? Quality assurance.

Manual quality inspection can impact your operating costs in more ways than one. An operator can only inspect one feature at a time, making quality control a time-intensive process on an otherwise fast-moving production line – meaning some features get missed, or fewer parts are completed. Errors are a susceptibility with repetitive inspection; missed defects can result in insufficient parts moving through the manufacturing process, causing later losses either through rework or replacement. When inspections are limited by human capacity, fewer features are reviewed and defects slip through the cracks – slowing down the process and costing you money in the long run.

Adding AI Vision to the mix solves those problems, keeping operations running smoothly and costs predictable. Jupiter AI Vision software tackles complex inspection, combining powerful software with a custom hardware stack to ensure rapid, accurate results in real-time without fail. It can work around the clock without a break, keeping pace with production as parts move through the line. Jupiter can read multiple features at once, providing an immediate pass/fail result relative to your own rejection criteria – and then locally stores the data and syncs straight to your dashboard or PLC so you retain complete control over your data. At the same time, each scan is inferred for anomalies so downstream models learn from the most accurate, up-to-date data directly from your own product – no defaults, no averages, and no hallucinations. It’s technology that integrates into your existing system, keeping your workflow familiar while boosting output and freeing up machine operators for more complex tasks.  

Functionality

Jupiter software can be used to inspect in both 2D and 3D, making it a versatile addition to your shop’s tech landscape and highly customizable to your product, line, and operation. The 2D application is best suited for weld inspections, running custom deep learning models on 2D cameras – no specialty optics, just 2D cameras and powerful software ensuring each bead matches predefined acceptance criteria. Operating with 99.4% detection accuracy, Jupiter can pinpoint defects across 8+ classes – and stream pass/fail results directly to the PLC to reject, rework, or pass the part. The 3D capability offers a full inspection of every blemish, scratch, or dent in each part Jupiter analyzes with micron-level precision.

How it Works

First, cameras mounted to custom hardware configurations – whether they are robots, linear slides, or turn-tables – scan the part with a full 360 halo view. Next, Jupiter generates a full point cloud, allowing you to review every aspect of your part in extreme detail all at once rather than one or two features at a time. Finally, once the full point cloud has been generated, our industry-leading AI algorithms determine which features fail against your quality standards. From start to finish, the process lasts only seconds – Jupiter works quickly enough to inspect every single part that goes through the line, meaning defects won’t slip through the cracks.

With a combination of advanced AI vision and sophisticated algorithms, the Jupiter software boosts defect detection capabilities, ensuring each and every part meets your standards – and fewer errors, detected within seconds in time with your production line, means real impact to your bottom line.