EVT Scratch Inspector – detection of surface defects with Machine Learning

With the intelligent, on Machine Learning based EVT Scratch Inspector, the detection of defects on surfaces is very easy. Based on a Deep Learning method, the Scratch Inspector detects any distrubances on the surface, similar to the human eye. Thanks to the Deep Learning base, the setup of the software is also very easy.

With only one image and simple illumination

Also the illumination has become a lot easier to handle. Whereas it is necessary in present common methods that the component part remains still for a moment during several images are captured, as this method is based on scenic illumination, the EVT ML Scratch Inspector only needs one image to detect the defect. This means that the components part does not have to be kept still and no elaborate illumination is necessary. A simple diffuse light is sufficient to detect scratches on a surface. As standard the Scratch Inspector is availble in three different illumination variations. Depending on the requirements, the illuminations can be easily adapted to the different sizes of the component parts.

Easily integrated into the plant and ready-to-use

The EVT Scratch Inspector is ready-to-use. The smart system has everything to integrate it directly into a plant. It has 8 in- and 8 outputs in 24 V technology to directly control a plant or to communicate with the PLC and the system is also equipped with a Profinet option. Therefore it is sufficient to connect the EVT Scratch Inspector to the PLC.

The system has the appropriate LED outputs to connect the illumination directly to the Scratch Inspector. The 24V power supply guarantees, that the Inspector can be run on a processor power supply.

Thanks to the sophisticated GUI it is possible to easily define test procedures. The algorithm is based on Machine Learning and takes almost the complete configuration of the user. The plant is therefore ready for the inspection in only a few mouse clicks.

Industry 4.0 compatible

The errors are either reported directly via I/O signal or an ejector is activated. Furthermore information can be sent via Profinet interface to the system control. Therefore the smart sensor is easily integrated into an Industry 4.0 environment. A data base connection and a detailed statistic allows the user to directly display the detected errors.

The system can be run as stand-alone system as well as headless sensor, which can be remotely programmed as OIP (Optical Inspectin Processor) and integrated into the plant.

EyeMulti-Inspect complete package: camera, processor, illumination, evaluation software

A completely ready-to-use alternative is the EyeMulti-Inspector. This system is a compact setup, where everything is integrated: camera hardware, processor and illumination. It can be positioned directly over the inspection location and all adjustments for camera, lens and illumination are included in the system.

Additionally, a tracker unit can eject the defect component parts at a defined position. To achieve that, the system has an incremental encoder input and several sorting outputs. So different quality classes can be created. The EyeMulti-Inspect is equipped with the same software as the Scratch Inspector. Thanks to a variance in illuminations it is also usable for other applications.

Spread the love

One thought on “EVT Scratch Inspector – detection of surface defects with Machine Learning

Leave a Reply

Your email address will not be published. Required fields are marked *