labelme标注不同物体显示不同颜色

labelme标注不同类别物体显示不同的颜色

1、labelme的版本

2、下载labelme后进行文件修改

由于博主想要的是rgb三通道的彩色图，所以标注出的是三通道的rgb值。
因此本教程适合想要标注的物体是彩色图案的读者，不是灰度图。

2.1 label.py文件的修改

#博主的参考路径为：
C:\ProgramData\Anaconda3\envs\labelme\Lib\site-packages\imgviz

修改如下内容：

import numpy as np

from . import color as color_module
from . import draw as draw_module


def label_colormap(n_label=256, value=None):
    """Label colormap.

    Parameters
    ----------
    n_labels: int
        Number of labels (default: 256).
    value: float or int
        Value scale or value of label color in HSV space.

    Returns
    -------
    cmap: numpy.ndarray, (N, 3), numpy.uint8
        Label id to colormap.

    """

    def bitget(byteval, idx):
        return (byteval & (1 << idx)) != 0

    cmap = np.zeros((n_label, 3), dtype=np.uint8)
    for i in range(0, n_label):
        id = i
        r, g, b = 0, 0, 0
        for j in range(0, 8):
            r = np.bitwise_or(r, (bitget(id, 0) << 7 - j))
            g = np.bitwise_or(g, (bitget(id, 1) << 7 - j))
            b = np.bitwise_or(b, (bitget(id, 2) << 7 - j))
            id = id >> 3
        cmap[i, 0] = r
        cmap[i, 1] = g
        cmap[i, 2] = b
        
    #我认为添加的一段，自己设置不同类别的rgb值
    cmap[0,:] = [255,255,255] #背景的rgb值
    cmap[1,:] = [255,0,0] #车的rgb
    cmap[2,:] = [0,255,0] #人的rgb

    if value is not None:
        hsv = color_module.rgb2hsv(cmap.reshape(1, -1, 3))
        if isinstance(value, float):
            hsv[:, 1:, 2] = hsv[:, 1:, 2].astype(float) * value
        else:
            assert isinstance(value, int)
            hsv[:, 1:, 2] = value
        cmap = color_module.hsv2rgb(hsv).reshape(-1, 3)
    return cmap


def label2rgb(
    label,
    img=None,
    alpha=0.5,
    label_names=None,
    font_size=30,
    thresh_suppress=0,
    colormap=None,
    loc="centroid",
):
    """Convert label to rgb.

    Parameters
    ----------
    label: numpy.ndarray, (H, W), int
        Label image.
    img: numpy.ndarray, (H, W, 3), numpy.uint8
        RGB image.
    alpha: float
        Alpha of RGB (default: 0.5).
    label_names: list of string
        Label id to label name.
    font_size: int
        Font size (default: 30).
    thresh_suppress: float
        Threshold of label ratio in the label image.
    colormap: numpy.ndarray, (M, 3), numpy.uint8
        Label id to color.
        By default, :func:`~imgviz.label_colormap` is used.
    loc: string
        Location of legend (default: 'centroid').
        'lt' and 'rb' are supported.

    Returns
    -------
    res: numpy.ndarray, (H, W, 3), numpy.uint8
        Visualized image.

    """
    if colormap is None:
        colormap = label_colormap()

    res = colormap[label]

    random_state = np.random.RandomState(seed=1234)

    mask_unlabeled = label < 0
    res[mask_unlabeled] = random_state.rand(*(mask_unlabeled.sum(), 3)) * 255

    if img is not None:
        if img.ndim == 2:
            img = color_module.gray2rgb(img)
        res = (1 - alpha) * img.astype(float) + alpha * res.astype(float)
        res = np.clip(res.round(), 0, 255).astype(np.uint8)

    if label_names is None:
        return res

    if loc == "centroid":
        for label_i in np.unique(label):
            if label_i == -1:
                continue  # unlabeled

            mask = label == label_i
            if 1.0 * mask.sum() / mask.size < thresh_suppress:
                continue
            y, x = np.array(_center_of_mass(mask), dtype=int)

            if label[y, x] != label_i:
                Y, X = np.where(mask)
                point_index = np.random.randint(0, len(Y))
                y, x = Y[point_index], X[point_index]

            text = label_names[label_i]
            height, width = draw_module.text_size(text, size=font_size)
            color = color_module.get_fg_color(res[y, x])
            res = draw_module.text(
                res,
                yx=(y - height // 2, x - width // 2),
                text=text,
                color=color,
                size=font_size,
            )
    elif loc in ["rb", "lt"]:
        unique_labels = np.unique(label)
        unique_labels = unique_labels[unique_labels != -1]
        text_sizes = np.array(
            [
                draw_module.text_size(label_names[l], font_size)
                for l in unique_labels
            ]
        )
        text_height, text_width = text_sizes.max(axis=0)
        legend_height = text_height * len(unique_labels) + 5
        legend_width = text_width + 40

        height, width = label.shape[:2]
        legend = np.zeros((height, width, 3), dtype=np.uint8)
        if loc == "rb":
            aabb2 = np.array([height - 5, width - 5], dtype=float)
            aabb1 = aabb2 - (legend_height, legend_width)
        elif loc == "lt":
            aabb1 = np.array([5, 5], dtype=float)
            aabb2 = aabb1 + (legend_height, legend_width)
        else:
            raise ValueError("unexpected loc: {}".format(loc))
        legend = draw_module.rectangle(
            legend, aabb1, aabb2, fill=(255, 255, 255)
        )

        alpha = 0.5
        y1, x1 = aabb1.round().astype(int)
        y2, x2 = aabb2.round().astype(int)
        res[y1:y2, x1:x2] = (
            alpha * res[y1:y2, x1:x2] + alpha * legend[y1:y2, x1:x2]
        )

        for i, l in enumerate(unique_labels):
            box_aabb1 = aabb1 + (i * text_height + 5, 5)
            box_aabb2 = box_aabb1 + (text_height - 10, 20)
            res = draw_module.rectangle(
                res, aabb1=box_aabb1, aabb2=box_aabb2, fill=colormap[l]
            )
            res = draw_module.text(
                res,
                yx=aabb1 + (i * text_height, 30),
                text=label_names[l],
                size=font_size,
            )
    else:
        raise ValueError("unsupported loc: {}".format(loc))

    return res


def _center_of_mass(mask):
    assert mask.ndim == 2 and mask.dtype == bool
    mask = 1.0 * mask / mask.sum()
    dx = np.sum(mask, 0)
    dy = np.sum(mask, 1)
    cx = np.sum(dx * np.arange(mask.shape[1]))
    cy = np.sum(dy * np.arange(mask.shape[0]))
    return cy, cx

需要修改的地方博主已经做了注释和说明。

2.2 json_to_dataset.py文件的修改

#博主的参考路径：
C:\ProgramData\Anaconda3\envs\labelme\Lib\site-packages\labelme\cli

修改如下内容：

import argparse
import base64
import json
import os
import os.path as osp

import imgviz
import PIL.Image

from labelme.logger import logger
from labelme import utils


def main():
    logger.warning('This script is aimed to demonstrate how to convert the'
                   'JSON file to a single image dataset, and not to handle'
                   'multiple JSON files to generate a real-use dataset.')

    parser = argparse.ArgumentParser()
    parser.add_argument('json_file')
    parser.add_argument('-o', '--out', default=None)
    args = parser.parse_args()

    json_file = args.json_file

    if args.out is None:
        out_dir = osp.basename(json_file).replace('.', '_')
        out_dir = osp.join(osp.dirname(json_file), out_dir)
    else:
        out_dir = args.out
    if not osp.exists(out_dir):
        os.mkdir(out_dir)

    data = json.load(open(json_file))
    imageData = data.get('imageData')

    if not imageData:
        imagePath = os.path.join(os.path.dirname(json_file), data['imagePath'])
        with open(imagePath, 'rb') as f:
            imageData = f.read()
            imageData = base64.b64encode(imageData).decode('utf-8')
    img = utils.img_b64_to_arr(imageData)

    label_name_to_value = {'_background_': 0,'car':1,'person':2}  #在这一行更改，改成自己的类别数
    for shape in sorted(data['shapes'], key=lambda x: x['label']):
        label_name = shape['label']
        if label_name in label_name_to_value:
            label_value = label_name_to_value[label_name]
        else:
            label_value = len(label_name_to_value)
            label_name_to_value[label_name] = label_value
    lbl, _ = utils.shapes_to_label(
        img.shape, data['shapes'], label_name_to_value
    )

    label_names = [None] * (max(label_name_to_value.values()) + 1)
    for name, value in label_name_to_value.items():
        label_names[value] = name

    lbl_viz = imgviz.label2rgb(
        label=lbl, img=imgviz.asgray(img), label_names=label_names, loc='rb'
    )

    PIL.Image.fromarray(img).save(osp.join(out_dir, 'img.png'))
    utils.lblsave(osp.join(out_dir, 'label.png'), lbl)
    PIL.Image.fromarray(lbl_viz).save(osp.join(out_dir, 'label_viz.png'))

    with open(osp.join(out_dir, 'label_names.txt'), 'w') as f:
        for lbl_name in label_names:
            f.write(lbl_name + '\n')

    logger.info('Saved to: {}'.format(out_dir))


if __name__ == '__main__':
    main()

文中需要修改的地方博主已经进行了说明。

3、如何使用修改后的文件

label.py和json_to_dataset.py修改后不用管。
当labelme标注完自己的数据集后，会生成对应的.json文件，此时需要将文件进行批量化生成json文件夹，里面包含了原图，标签，txt等文件。博主弄了一个批量化生成json文件夹的python文件。
此文件博主已经上传到csdn,读者可以免积分下载。
链接：https://download.csdn.net/download/weixin_44753371/12298864

import os
path = 'D:/Image_Segmentic/Semantic-Segmentation-master/dataset2/label'  # path为json文件存放的路径
json_file = os.listdir(path)
os.system("activate labelme") # 博主labelme所在的环境名就叫labelme，读者应修改成activate [自己labelme所在的环境名]
for file in json_file: 
    os.system("labelme_json_to_dataset.exe %s"%(path + '/' + file))

使用方法：

在虚拟空间建立labelme，要进入到labelme的虚拟空间。尽管会提示警告，但可以进行批量化处理，因为labelme的本意是演示单个文件的标注，所以本身不支持批量化处理。

本文的参考博客：
1
2