Our plant quality control systems with visual inspection exploit the advantages offered by artificial intelligence, providing companies with a solution that solves complex industrial problems, controlling quality during different production phases based on the type of product and characteristics of the object under analysis.

Automated systems are capable of verifying the entire production process in real time — much to the detriment of traditional processes which are often linked to subjective and inefficient manual methods.

Thanks to the skills of our specialized technicians, we provide a turnkey system equipped with the most advanced AI-based visual inspection techniques that accurately analyze images and data in real time.

This innovative system is simple to train and validate.

Because the model is trained using compliant pieces, it identifies anomalies deviating from suitability parts.

Our application identifies a wide variety of product defects:

  • Check differences
  • Presence & absence
  • Positioning
  • Surface and appearance 
  • Alignment

Anomaly Detection

Our AI quality control system with machine vision is based on anomaly detection, which consists of examining specific data points and events that are suspicious because they differ from the established behavior model.

Developing an AI model begins with training activities based on classifying the objects to be verified during quality control and loading a small number of images necessary for that training.

Our solution allows you to solve industrial problems through two methods: classification and segmentation.

Classification

Classification mode is based on identifying defects via a heat map, which allows you to assign an initial binary “good/fail” or multi-class value to the images, identifying images belonging to different defect categories.

Segmentation 

Segmentation locates and classifies anomalies, previously introduced during the training phase, within the image, returning a mask defining the area.

Main features

  • Pre-trained model, simple to train and validate using few examples (~10 images)
  • Templates are managed through a single interface
  • Developed for industrial environments
  • Customizable dashboards
  • Configurable alerts
  • Solution available on cloud or edge
  • System using the most well-known machine vision brands
  • Industry 5.0

Advantages

  • Improved product quality
  • Increased production performance
  • Reduction of mirrors
  • Greater accuracy in product control (>95%)
  • Increased customer loyalty
  • Improved company image