将灰度图像转换为 RGB，反之亦然

Question

I wrote a code to convert ground truth grayscale mask to RGB, and Vice Versa, but RGB2Grayscales does not work as expected?
Example of label, and converted RGB

from __future__ import print_function, absolute_import, division
from collections import namedtuple
import numpy as np
import cv2

Label = namedtuple('Label',
                   ['name', 'id', 'trainId', 'category', 'categoryId', 'hasInstances', 'ignoreInEval', 'color', ])

labels = [
    #       name                     id    trainId   category            catId     hasInstances   ignoreInEval   color
    Label('unlabeled', 0, 19, 'void', 0, False, True, (0, 0, 0)),
    Label('ego vehicle', 1, 19, 'void', 0, False, True, (0, 0, 0)),
    Label('rectification border', 2, 19, 'void', 0, False, True, (0, 0, 0)),
    Label('out of roi', 3, 19, 'void', 0, False, True, (0, 0, 0)),
    Label('static', 4, 19, 'void', 0, False, True, (0, 0, 0)),
    Label('dynamic', 5, 19, 'void', 0, False, True, (111, 74, 0)),
    Label('ground', 6, 19, 'void', 0, False, True, (81, 0, 81)),
    Label('road', 7, 0, 'flat', 1, False, False, (128, 64, 128)),
    Label('sidewalk', 8, 1, 'flat', 1, False, False, (244, 35, 232)),
    Label('parking', 9, 19, 'flat', 1, False, True, (250, 170, 160)),
    Label('rail track', 10, 19, 'flat', 1, False, True, (230, 150, 140)),
    Label('building', 11, 2, 'construction', 2, False, False, (70, 70, 70)),
    Label('wall', 12, 3, 'construction', 2, False, False, (102, 102, 156)),
    Label('fence', 13, 4, 'construction', 2, False, False, (190, 153, 153)),
    Label('guard rail', 14, 19, 'construction', 2, False, True, (180, 165, 180)),
    Label('bridge', 15, 19, 'construction', 2, False, True, (150, 100, 100)),
    Label('tunnel', 16, 19, 'construction', 2, False, True, (150, 120, 90)),
    Label('pole', 17, 5, 'object', 3, False, False, (153, 153, 153)),
    Label('polegroup', 18, 19, 'object', 3, False, True, (153, 153, 153)),
    Label('traffic light', 19, 6, 'object', 3, False, False, (250, 170, 30)),
    Label('traffic sign', 20, 7, 'object', 3, False, False, (220, 220, 0)),
    Label('vegetation', 21, 8, 'nature', 4, False, False, (107, 142, 35)),
    Label('terrain', 22, 9, 'nature', 4, False, False, (152, 251, 152)),
    Label('sky', 23, 10, 'sky', 5, False, False, (70, 130, 180)),
    Label('person', 24, 11, 'human', 6, True, False, (220, 20, 60)),
    Label('rider', 25, 12, 'human', 6, True, False, (255, 0, 0)),
    Label('car', 26, 13, 'vehicle', 7, True, False, (0, 0, 142)),
    Label('truck', 27, 14, 'vehicle', 7, True, False, (0, 0, 70)),
    Label('bus', 28, 15, 'vehicle', 7, True, False, (0, 60, 100)),
    Label('caravan', 29, 19, 'vehicle', 7, True, True, (0, 0, 90)),
    Label('trailer', 30, 19, 'vehicle', 7, True, True, (0, 0, 110)),
    Label('train', 31, 16, 'vehicle', 7, True, False, (0, 80, 100)),
    Label('motorcycle', 32, 17, 'vehicle', 7, True, False, (0, 0, 230)),
    Label('bicycle', 33, 18, 'vehicle', 7, True, False, (119, 11, 32)),
    Label('license plate', -1, -1, 'vehicle', 7, False, True, (0, 0, 142)),
]


def trainIdToColor(trainId: int):
    for l in labels:
        if l.trainId == trainId:
            color = l.color
            break
    return color


def colortoTrainId(rgbColor):
    trainId = 0
    for l in labels:
        if l.color == rgbColor:
            trainId = l.trainId
            break
    return trainId


def gray2color(grayImage:np.ndarray,num_class:int):
    rgbImage=np.zeros((grayImage.shape[0],grayImage.shape[1],3),dtype='uint8')
    for cls in range(num_class):
        row,col=np.where(grayImage==cls)
        if (len(row)==0):
            continue
        color=trainIdToColor(cls)
        rgbImage[row,col]=color
    return rgbImage


def color2gray(colorImage:np.ndarray, bgr_color_space:bool):
    if bgr_color_space:
        colorImage = cv2.cvtColor(colorImage, cv2.COLOR_BGR2RGB)


    unique_color=np.unique(colorImage.reshape(-1, colorImage.shape[2]), axis=0)
    gray=np.zeros((colorImage.shape[0],colorImage.shape[1]),dtype=np.float32)
    for uc in unique_color:

        where_cond1= np.logical_and(colorImage[:,:,0]==uc[0],
                                    colorImage[:,:,1]==uc[1],
                                    colorImage[:,:,2]==uc[2])
        row,col=np.where(where_cond1)
        gray[row,col]=colortoTrainId(tuple(uc))

    return gray

when I use gray2color Everythings works fine. but when I try to convert an RGB image via color2gray , it converts but the result is not the same as original gray image. (changes 19s to 13, other classes are fine) . I checked to code multiple times but I don't know why I got bad results.
for clarification
as you can see in returned grayscale there's no 19 value,all added up to 13s.

original grayscale:
unique: 0   1   2   4   5   7   8   10  11  13  19
count:  624649  168701  819940  2802    24885   12192   42082   37098   6791    115270  242742

returned grayscale:
unique: 0   1   2   4   5   7   8   10  11  13
count:  624649  168701  819940  2802    24885   12192   42082   37098   6791    358012

also color2gray function is very slow and time-consuming!

Answer 1

your trainId value is too low that's why your grayscale image is very dark. Increase the value of trainId in the Labels list and make your gray image datatype to be uint8

def color2gray(colorImage:np.ndarray, bgr_color_space:bool):
    if bgr_color_space:
        colorImage = cv2.cvtColor(colorImage, cv2.COLOR_BGR2RGB)


    gray=np.zeros((colorImage.shape[0],colorImage.shape[1]),dtype=np.uint8)

    for row in range(colorImage.shape[0]):
        for col in range(colorImage.shape[1]):
            # either multiply the result of colortoTrainId(tuple(uc)) with some number
            # or change the value of trainId in the Labels list
            gray[row,col]=colortoTrainId(tuple(colorImage[row,col]))*10

    return gray

Answer 2

I found the answer of 1 question myself

np.logical_and() just take two arrays

so it must be used nested like this

where_cond1=np.logical_and(np.logical_and(colorImage[:,:,0]==uc[0],colorImage[:,:,1]==uc[1]), colorImage[:,:,2]==uc[2])

but the speed problem remains, any idea how to vectorize assignment?