AI Machine Vision for manufacturers
We help manufacturers automate complex quality control and production processes with cutting-edge AI & Vision solutions. To guarantee seamless integration for end-to-end solutions, we collaborate closely with hardware and machinery partners.
Trusted by leading manufacturers in Europe
AI Machine Vision
At ML6, we don't just automate quality inspection; we use AI to optimize complex quality control processes and apply quality insights to improve other processes and drastically improve production efficiency.
With over a decade of expertise in computer vision, we deliver the most accurate results for our customers through advanced machine learning models.
The benefits of using ML for machine vision include:
Model assisted data labelling with auto annotations
Full defect detection traceability & transparency
Proficient in complex object processing and quality parameters
No constant retraining for product variations
Scalable & robust
Visual Inspection AI
Flexible and accurate visual quality inspection across your whole production line (raw material sourcing, manufacturing, packaging) to get the most accurate quality insights for complex and dynamic products, in real-time.
Automated defect removal
In many cases, complex products still need manual defect removal, even with quality grading. We offer operators assistance by automating some tasks or having them focus solely on validating our automated defect removals.
Machine process steering
Getting insights into detailed aspects of production, such as defects at various stages, allows for swift adjustments to machine settings like heating times and temperatures. We can assist in leveraging this data to enhance quality, speed, and efficiency in production with AI digital twins.
Use case examples
Here are some complex use cases we've solved for clients in the following industries:
Textile
Detecting rug edge lines with our AI model, enabling precise folding despite texture, pattern, and colour variations, enhancing production efficiency.
Metal
Splitting microscopic SEM images into multiple parts to classify phosphate coverage on steel wire, leveraging quantitative metrics to adjust production parameters automatically.
Wood
Identifying wooden plank defects, halting production for critical issues caused by machine failure. Saving time and costs, reducing waste, improving quality and providing machine improvement insights.
Pharma
Developing an application using hyperspectral sensing for lab scientists to detect and classify vial defects, achieving 98.4% accuracy.
Food
Using cameras and depth sensors with an algorithm to identify and reject defect pallets, preventing downtime in the packaging pipeline and avoiding safety risks, export delays, and fines.
How AI/ML makes the difference
Traditional methods and machine vision struggle with complex defects and underutilise quality insights. Our AI solutions offer an economical way to deliver value fast, while tailoring to your needs where customization is essential.
Implementing AI (Machine Vision) offers many benefits including:
Minimized production waste and machine downtime
More consistent product quality
Enhanced quality inspection efficiency and accuracy
Higher production throughput
Data-driven continuous improvement cycle
Success Stories
Transforming Wienerberger’s Quality Control with AI
In collaboration with Cosmec, we’ve deployed a system that proactively detects and removes defect bricks, considering complex natural product fluctuations, to enhance production efficiency, reduce costs, minimize waste, and improve product quality.
Deploying AI algorithms on a multispectral camera for CNHi
We’ve implemented an object segmentation algorithm capable of detecting broken rice, sticks, needles and leaves. With the purpose to automate waste detection in rice, adjusting combiner settings accordingly. This increases harvest amount and quality for farmers, providing CNHi with a significant competitive edge.
Optimizing Balta’s rug production with AI-driven machine steering
We’ve been helping Balta with multiple automaton projects, with carpet folding automation being one of them. The system is able to detect rug edge lines with our Vision AI model, enabling precise folding despite texture, pattern, and colour variations, enhancing production efficiency.
Improving production processes with computer vision
Over the past few years, ML6 has collaborated with Unilin to help their team to overcome complex computer vision challenges, achieving pixel-perfect accuracy and robustness across different product types. Hear more about the project from Michiel Van Acker (Unilin) as he explains the incredible impact of artificial intelligence in improving their production processes.
Implementing & Integrating the Solution
Our unique approach results in a tailored solution based on modular building blocks. This way, we offer the best of both worlds: we customise where needed and provide the cost-effectiveness and scalability of a product.
Capture
We provide the hardware to capture the images of your products, fitted to your needs. We install and monitor them and guarantee their operations.
Inspect
Our ML core processes the captured images using a custom model to detect or classify the products according to your use case.
Tailored
Training
We train a custom machine learning model on your data to meet all your quality inspection needs.
Integrate
We integrate with your PLC’s to communicate the inspection results and send steering instructions.
Tailored
PLC integration
We build a bespoke integration with your machines to communicate, intervene and steer.
Report
We show production metrics on standard reports and dashboards.
Tailored
Labelling and retraining
We expose a custom labelling and retraining setup to support a continuous improvement cycle for your quality inspection process
Maintain
We provide maintenance and support for the solution to guarantee security patches and future functionality upgrades.
Implementation Timeline
We swiftly deliver value with a pre-FAT consultation, FAT approval, hardware integration, 3-4 weeks of data collection, ML model training, a two-week pilot, and SAT within three months, with acceptance criteria outlined beforehand.
