Augmentation Glossary

Horizontal flip & Vertical flip

The horizontal and vertical flip augmentations mirror the images horizontally or vertically and add the modified images to the training set in order to improve accuracy and robustness. 

These augmentations are recommended for situations where the orientation of the inspected products can change but it's not the goal of the inspection to check the orientation (e.g. inspection of fabric/wood/metallic surface, products on a conveyor belt, etc.).

On the other hand, it's not suitable for inspecting defects on static visual inspection, such as PCB (Printed Circuit Boards), where the orientation of the parts is being checked and we e.g. want to recognize if some individual part is flipped. In this case, the flip augmentations should be disabled for the model to learn the different classes correctly.

Open

Open

Width shift (min 0, max 100) [%]

A shift to the left or right will occur.

Height shift (min 0, max 100) [%]

A shift up or down will occur.

Color jittering (min 0, max 50)

Color jittering augmentation adjusts the color of RGB image, with a selected value. This augmentation may be suitable for cases where images have a wide range of colors or shades.

It has been effectively used in previous cases, where for example, the image’s background was very dark and we were attempting to detect holes for a target part.

Brightness resistance (min 0, max 100)

The name itself is suggestive, it’s a value to alter the resistance to brightness within the picture, in other words, it’s the resistance to changes in lighting.

Brightness Resistance values should be increased if you think the lighting is affecting the software’s ability to identify defects due to the brightness/lightness changes on the images.

Saturation resistance (min 0, max 100)

Saturation is the strength or purity of the color and represents the amount of gray in proportion to the hue. A "saturated" color is pure and an "unsaturated" color has a large percentage of gray.

Contrast resistance (min 0, max 100)

Contrast is the difference in brightness, color, or tone between the darkest and lightest parts of an image, which affects the visual separation and distinction between different elements or regions within the image.

This augmentation creates images with increased/decreased contrast. By selecting a larger number you can enlarge the possible range of contrast increase/decrease. The model should then be more resistant to changes in image contrast.

Shear (min 0, max 360)

The object can be skewed by a given factor in the direction of x or y axis.