Artificial intelligence and machine vision
Intelligent machine vision for smart production processes
Artificial intelligence is revolutionizing machine vision by adding AI-based approaches to the efficiency and precision of inspection and monitoring processes in a meaningful way. At aku, we use state-of-the-art AI technologies to fulfill your machine vision requirements.
Our AI solutions
Fulfilling your machine vision requirements with the latest technology
Cognex VisionPro Deep Learning
Powerful AI-based image analysis software for complex applications
Cognex ViDi
High-performance tools for repetitive AI-based tasks
Smart cameras & Edge Learning
Powerful hardware with AI algorithms directly inside the camera
Halcon AI
Efficient training of high-performance networks thanks to Halcon AI
ONNX-Worker
Training of own neural networks and integration in aku.visionManager®
Advanced deep learning technologies for your production
Classification
During classification, the entire image is assigned to a class. Therefore, only the corresponding sample images need to be made known to the network. In this process, each image is assigned an associated confidence value for the class. This indicates the match rate of the presented image with the respective class.
Object detection
Object detection enables the detection and spatial localization of various trained classes in an image. A recognized object is assigned to a class and outlined in a rectangle. The recognition and differentiation of different objects is also used for reading fonts and for optical character recognition (OCR).
Segmentation
Segmentation represents a refinement of object detection. It is a more precise version of object detection. In this method, each individual image pixel, including the background, is assigned a class. This allows a detected object to be precisely segmented.
Anomaly detection
Anomaly detection identifies deviations from a good version. To train the system, defect-free objects are used first. The system then recognizes errors (e.g. cracks, holes, scratches) in the image as deviations. The advantage is that the system learns to find errors without knowing them beforehand.
Case studies
You might also be interested in:
