AI-Powered Visual Defect Detection
Adaptive AI driving the Next-Generation of Visual Inspection Solutions
Vanti Visual Defect Detection is an AI-powered visual inspection solution that improves quality and efficiency in the manufacturing industry.
It uses adaptive AI to monitor and adapt to changes in the manufacturing environment. It also supports unsupervised learning for the classification of defect types and the discovery of new types.
The solution can be easily implemented in a matter of days, and offers real-time deployment options on-premises or in the cloud. It enables high-volume production lines to utilize AI for identifying sub-quality products.
Visual Inspection Challenges
Efficient and accurate visual inspection is crucial for maintaining product quality and production schedules in the manufacturing industry.
However, manual inspection methods fall short in meeting these needs:
New, AI-based visual inspection solutions still face significant challenges, such as:
Vanti’s Adaptive AI-based Visual Defect Detection Solution
Vanti is a cloud-based manufacturing optimization solution that includes built-in AI-driven applications. One of these applications is Visual Defect Detection.
Any manufacturing professional can use Vanti’s platform to upload data, train a model and deploy a solution within days.
Sample Use Cases
Automotive Paint Inspection:
Reduce manual labor by 50%
In the automotive paint process, small and hard-to detect defects are common. Vanti’s technology, in conjunction with cameras that scan the vehicle, automatically locates and categorizes defects around the vehicle, providing operators with a “defect map” to help them address the issues, saving the manufacturers time and money.
Predictive Quality in Electronics:
Increase efficiency by 5%
The manual assembly of electronics products is prone to errors, and issues are typically only detected at the end of the production line.
Vanti’s AI allows manufacturers to identify issues mid-production, saving significant resources and time, and improving throughput and product quality.
High-Volume CPG Inspection:
Increase throuput by 4%
In the consumer packaged goods industry, products are produced at a rapid rate. Vanti’s AI can detect issues with packaging or filling in real-time and when integrated with a sorting unit, can remove sub-quality items, increasing the overall quality and reducing inspection resources.
Vanti’s 3-Step Deployment Process
Setup: Vanti is compatible with a wide range of cameras, including still and video cameras. If a camera is already in place, Vanti can utilize existing images. In cases where new cameras or automation equipment need to be installed, Vanti works with its ecosystem partners to provide an all-inclusive solution.
Images upload: The user needs to have access to the images produced by the camera, and upload them into Vanti either from a computer or using one of Vanti’s connectors to AWS S3, Azure, etc.
Selecting Supervised or Unsupervised: Vanti supports both a supervised approach in which the user labels images as “good”, “bad” etc., or an unsupervised approach in which Vanti will cluster the images into clusters, enabling the user to process images that contain unknown issues
Training: With a click of a button, a new AI model is trained, normally in under 15
minutes, and the model predictions and model accuracy are displayed. If the results are not adequate, more images are uploaded and the model is retained until the required accuracy achieved.
Deployment: Using Vanti’s API, and optional edge deployment capability, a real-time stream of images is sent to Vanti, and a real-time stream of “predictions”, i.e. “good” or “bad” is provided. The prediction stream can be fed directly into a sorting arm, or viewed by an operator for further analysis and decision-making.