Visual quality control technology is key in assuring high‑quality manufactured goods and lowering the internal and external costs of failures.

Even present day solutions are predominantly based on traditional computer vision algorithms, which are often unable to deal with products with complex patterns, like fabrics and wood.

Every alteration in production calls for manual and time‑consuming reconfiguration, hindering the ability to quickly respond to the client’s needs.

Harnessing the power of recent developments in deep learning, deepsense.ai offers a solution that works well for complex patterns while eliminating the tedious manual configuration required by traditional computer vision algorithms. Our visual quality control solution is developed to best suit the specific client’s needs:

• The system’s architecture is designed to perform best in the client’s environment.
• As long as the goods produced are from the same domain, the system can be automatically reconfigured by merely presenting it with good examples. This is a key advantage whenever the defective products are rare or the anomalies are not reproducible.
• After a short learning phase, the system is ready to be used in production.
• The solution can be shipped as an all-in-one system addressing multiple use cases, including anomaly localization, defect classification and assessment of the severity and quantity of defects.

• Adaptability – it can be quickly retrained to perform new tasks.
• Self-learning – there’s no need for laborious manual fine-tuning of parameters.
• The ability to learn complex patterns – the characteristics of the products being manufactured are inferred automatically, so they need not be manually specified beforehand.