Automatic image classification with c-Alice
Software for process control in semiconductor manufacturing
For faster feedback loops, improved process control and enhanced product quality
Convanit provides a complete solution for AI based image classification to enable manufacturing companies to automatically analyze images of any type. Well suited for semiconductor industry, electronics manufacturing, automotive, plastics, pharmaceutical and food industry
Works for a large variety of industrial cameras, inspection and vision systems including optical and scanning electron microscopes, defect maps and electrical test data.
centralized
One centralized AI-framework for the whole production site
independent
Independent of imaging systems
customizable
optimized algorithms based on use case
"After a quick implementation into our infrastructure, we can use c-Alice to automatically evaluate images at many monitoring points in the manufacturing process. The application is easy to use, so engineers and technicians are able to implement their specific use cases themselves. By using c-Alice, we have already been able to identify several process problems at an early stage and save considerable costs by significantly improving yields."
Dejan Simic, Head of Metrology at Semikron-Danfoss
Your benefits
one for all
c-Alice can classify all types of images: any type of inspection / microscopy, wafer maps, test maps and others.
fast & accurate
c-Alice generates classification results in seconds. High accuracy is achieved by using fast and optimized algorithms.
complete workflow
c-Alice provides image and recipe management including versioning, monitoring, deployment and meta data integration. Low maintenance effort and reduced manpower ensure efficient classification in production.
easy set-up
c-Alice allows easy recipe set up without further AI knowledge. Users can manage images and context data comfortably in order to train, verify and release recipes.
flexible
c-Alice is based on a modern architecture and can be integrated flexibly in your production environment.
"With c-Alice, we are able to evaluate the data generated during our inspection processes more accurately and reliably. We can detect errors at different levels in the production process, we can see at which levels they occur and whether they are ultimately relevant to the product. Our experts can apply AI methods in a simple and intuitive way and transfer their knowledge to so-called recipes, which can be created, configured, trained, released and implemented using a simple workflow. We have a partnership with convanit and want to develop c-Alice together with them."
Site manager Jörg Amelung and defect density engineer Juliane Weise at Fraunhofer IPMS
Use Cases
Classification tasks
classification of failure types in defect images (scratches, holes, particles, stains, missing structures and more), reduction of pseudo defects, color control
Anomaly detection
find unexpected deviations
Pattern, structure and surface analysis
pattern recognition of defect and binsort wafer maps, structure analysis of materials and liquids and more
Object detection
counting, position control, completeness check
From classification towards characterization
Beyond just image classification, c-alice can process arbitrary data along with the images to characterize defects and other problems.
Who we are
10+ experts are working on the c-Alice project
convanit GmbH & Co. KG
combination with IT and data science
Founded in
in Germany
Experts with more than
years of industry expertise in production control
Projects in more than
manufacturing sites
Our Networks
Cooperation with several local AI expert groups within university and institutes
Silicon Saxony
VDI/GMM
SEF
Smart System Hub
Selection of our clients
Latest publications
KI zur Qualitätskontrolle von Platinen
24.02.2024
Limtronik geht nächste Schritte in der Automatisierung und in Richtung Künstliche Intelligenz. So optimiert die Elektronikfabrik die Qualitätskontrolle von Platinen mit einem Kamerasystem und automatisierter Bildauswertung. Zur Auswertung der Bilder kommt die Software c-Alice der Firma convanit mit KI-basierten Modellen zum Einsatz. Ziel ist es, den Verarbeitungsprozess abzusichern. Die Anwendung wurde am Standort in Limburg an der Lahn getestet und für die Dependance in Aurora (Colorado/USA) ausgeprägt.
Silicon Saxony - Smart Systems: Vom individuellen KI Anwendungsfall zur zeitnahen bezahlbaren Umsetzung in der Fertigung
Im KMU Bereich gibt es noch viele ‚ungehobene Schätze‘ im Bereich KI, wie zum Beispiel in der Fertigungskontrolle. Fachwissen, KI Methoden und Daten sind zwar vorhanden, jedoch fehlt für eine produktionsreife Lösung häufig noch das richtige Konzept zur pragmatischen Verbindung dieser Komponenten. Der Weg dahin muss nicht schwer sein, wenn man sich auf das Wesentliche konzentriert.
Digitale Welt: KI in der Fertigungskontrolle – Herausforderungen und Strategien
21.03.2023
Für die Bewertung von Zeitreihen-, Bild- und anderen Daten, ist gerade bei Zuordnungs- und Klassifizierungsaufgaben und der Erkennung von Mustern und Anomalien der Nutzen von KI Methoden hoch. Die dafür notwendige Rechenleistung ist heute ‚bezahlbar‘ und erprobte Mathematik sowie mächtige Open-Source Softwarebibliotheken stehen zur Verfügung. Dieser Artikel beschreibt Herausforderungen bei der Einführung von KI Methoden in der Fertigungskontrolle und die daraus abzuleitenden Strategien.
We are supported by
