Finally, let's take a quick look at the LBPH algorithm. Again, the process is similar. However, the algorithm factory takes the following optional parameters (in order):
- radius: The pixel distance between the neighbors that are used to calculate a cell's histogram (by default, 1)
- neighbors: The number of neighbors used to calculate a cell's histogram (by default, 8)
- grid_x: The number of cells into which the face is divided horizontally (by default, 8)
- grid_y: The number of cells into which the face is divided vertically (by default, 8)
- confidence: The confidence threshold (by default, the highest possible floating-point value so that no results are discarded)
With default arguments, the model declaration ...