: In robotics, 1x3 matrices are often used as translation vectors to denote the physical movement or position of an object in 3D space.
A 1x3 filter generates a "feature" by performing a —a mathematical operation where the filter slides across an input array to identify specific patterns like edges, gradients, or intensities. S W A T 1x3
: The filter "slides" to the right by a set distance (usually 1 pixel, known as a stride of 1) and repeats the calculation for the next set of inputs. : In robotics, 1x3 matrices are often used
: You define a filter with specific weights, such as [2, -1, 0.1] . : You define a filter with specific weights,
: Specialized filters can detect sharp changes in intensity, effectively "outlining" shapes within data.
: The final sequence of these output values forms a feature map , which represents the detected presence and strength of the target pattern across the entire input. Practical Applications
: In 2D images, a 1x3 filter specifically targets horizontal patterns, which can be useful for identifying textures or long horizontal lines. Real-Time Vision for Robot Swat-Juggling