AI Visual Counter

WHAT THIS IS

• Factbird AI Visual Counter is a counting solution that uses a camera and an AI model to identify and count products as they move through a production line.
• Unlike traditional counters that rely on beam breaks, proximity sensing, or product separation, the AI Visual Counter analyzes video images to detect and count products based on their appearance and movement.
• The solution is designed for production environments where standard sensors struggle to provide accurate counts.

WHY IT MATTERS

• Some production lines cannot be reliably monitored using conventional counting sensors due to product shape, positioning, surface characteristics, or flow patterns.
• The AI Visual Counter enables production tracking in these environments, providing accurate count data that can be used for OEE, downtime analysis, and performance monitoring.
• By capturing production data that would otherwise be unavailable or unreliable, teams gain better visibility into operational performance.

WHEN YOU WOULD USE THIS

• Use this when:
  • Products are not separated into clear batches or individual lanes.
  • Products flow in multiple lanes or irregular patterns.
  • Items appear at varying angles or orientations.
  • Product surfaces are reflective, glossy, transparent, or otherwise difficult for traditional sensors.
  • Physical or environmental constraints prevent reliable laser or beam-break counting.

HOW IT WORKS

• A camera captures images of products moving through a defined counting area.
• The camera connects to a Factbird EDGE device running an AI counting model.
• The AI model analyzes the video stream and identifies products based on visual characteristics.
• Counting data is sent to the Factbird cloud platform, where it becomes available for production monitoring, OEE calculations, downtime tracking, and performance analysis.
• Each AI Visual Counter is trained and validated using images and production data from the customer's specific products and production environment.

KEY TERMS / COMPONENTS

• AI Visual Counter:
  • A camera-based counting solution that uses artificial intelligence to detect and count products.
     
• Industrial Camera:
  • The image capture device used to monitor the counting area.
     
• Factbird EDGE:
  • The local processing device that runs the AI counting model and transmits data to Factbird.
     
• AI Counting Model:
  • A trained machine learning model that identifies and counts products within the video stream.
     
• Counting Zone:
  • The area within the camera's field of view where products are detected and counted.
     
• Validation:
  • The process of testing and refining the AI model using real production data to ensure accurate counting.

COMMON MISUNDERSTANDINGS

• The AI Visual Counter is not a plug-and-play sensor. Each installation requires model training and validation using customer-specific production data.
• The solution does not eliminate installation requirements. Camera placement, lighting conditions, and production flow still significantly impact performance.
• The AI model is not universally trained for all products. New products, formats, or major process changes may require additional validation or retraining.
• The system is intended for counting products and production flow, not for general machine vision inspection applications.