Automatic inspections and less customer complaints
We want to empower your company.
We teach machines to see.
Our solutions use artificial intelligence and deep learning to automate your visual or image-based exams.
Classification & Segmentation – Example: Cups
We teach machines to see and name things and components. Also interesting regions or parts can be recognized and marked precisely.
Anomaly Detection – Example: Cookies
What does a good product look like? We show computers examples of faultless objects and tolerable defects. In this way the system learns what something has to look like and detects deviations from it.
Counting – Example: Screws
A computer can be trained to recognize and count properties in images. This allows products or packaging, for example, to be checked for completeness.
Approximately since 2012 (ImageNet Challenge) Deep Learning methods are superior to classical methods of image processing (at that time 74% accuracy). Today, these accuracies can reach +97% and are even better than humans.
A computer never gets tired of routine tasks and does them objectively. Deep learning models can also be more robust against external influences (e.g. exposure) than classical image processing.
Classic image processing and machine vision must be calibrated at great expense. An artificial intelligence learns error images automatically and can already be trained with between 20 and 200 images.
Our deep learning models can be easily integrated into different machines and processes. This saves time and reduces development effort.
We give you new powers.
If machines can see better and faster than a human being, this gives your company more time for demanding tasks and saves costs.
A closer inspection prevents slippage. This keeps your customers satisfied, saves you money and you are ready for the next audit.
A computer can analyze multiple images simultaneously and combine multiple inspections. This enables you to check more parts and increase production.
The early detection of changes and errors is particularly critical in batch production. If an entire batch has to be disposed, repackaged or decanted, automatic control with artificial intelligence offers great potential.
Automatic weld analysis
Together with VISUS Industry IT and Applus+ RTD, we are developing a software to analyze imperfections in X-ray images of welds.