Function densecrf

examples/demo_densecrf.py:6–24

input parameters: I : a numpy array of shape [H, W, C], where C should be 3. type of I should be np.uint8, and the values are in [0, 255] P : a probability map of shape [H, W, L], where L is the number of classes type of P should be np.float32

(I, P, param)

import matplotlib.pyplot as plt

def densecrf(I, P, param):
    """
    input parameters:
        I    : a numpy array of shape [H, W, C], where C should be 3.
               type of I should be np.uint8, and the values are in [0, 255]
        P    : a probability map of shape [H, W, L], where L is the number of classes
               type of P should be np.float32
        param: a tuple giving parameters of CRF (w1, alpha, beta, w2, gamma, it), where
                w1    :   weight of bilateral term, e.g. 10.0
                alpha :   spatial distance std, e.g., 80
                beta  :   rgb value std, e.g., 15
                w2    :   weight of spatial term, e.g., 3.0
                gamma :   spatial distance std for spatial term, e.g., 3
                it    :   iteration number, e.g., 5
    output parameters:
        out  : a numpy array of shape [H, W], where pixel values represent class indices. 
    """
    out = denseCRF.densecrf(I, P, param) 
    return out   

def convert_label_to_probability_map(label, color_list):
    [H, W, _] = label.shape