Detection
Detection engines are responsible for locating objects in the 3D scene captured by the Pickit camera. Different applications require different detection engines depending on the shape and arrangement of the objects. In the Applications section, you can learn more about how they are supported by Pickit, and which detection engines are available for each application.
Pickit has two general detection engines, each optimized for a different type of shape.
Teach is the most versatile engine, and well suited for most shapes of known fixed sizes, both simple and complex. It is the recommended detection engine for most applications, ranging from complex shapes to simple shapes (like cylinders) of known dimensions.
Flex is meant for detecting geometric shapes in 3D (cylinders and spheres) and 2D (squares, rectangles, circles and ellipses). It can detect instances of the same shape with similar or different sizes. Combined with AI segmentation feature, Flex can detect entire categories of objects with similar shapes, even if they are not known in advance. When detecting cylinders of known dimensions, Teach with cylinder models is the recommended engine.
Pickit also offers detection engines optimized for specific part shapes and arrangements:
Flex Depal, designed for depalletization applications, where the parts to pick are mostly flat and stacked in layers. It can reliably detect boxes (mixed and unmixed), bags and other palletizable items.
DeepAL, for unmixed depalletization applications. For new applications, we recommend using Flex Depal, as it is more versatile and easier to configure.
Bags, designed for detecting bags on a pallet. For new applications, we recommend using Flex Depal, as it is more versatile and easier to configure.
Pattern, which is similar to Flex. It looks for 2D shapes (rectangles, squares, circles and ellipses) of known fixed size, and is especially useful for detecting parts that are aligned and touching.
Hole, for detecting circular holes of known size on a surface.