官方链接: https://gluon-cv.mxnet.io/build/examples_datasets/detection_custom.html#sphx-glr-download-build-examples-datasets-detection-custom-py

1. 官方demo

from gluoncv.data import VOCDetection
class VOCLike(VOCDetection):
    CLASSES = ['person', 'dog']
    def __init__(self, root, splits, transform=None, index_map=None, preload_label=True):
        super(VOCLike, self).__init__(root, splits, transform, index_map, preload_label)

dataset = VOCLike(root='VOCtemplate', splits=((2018, 'train'),))
print('length of dataset:', len(dataset))
print('label example:')
print(dataset[0][1])

在官方demo的基础上增加了显示

from gluoncv import utils
import mxnet as mx
import numpy as np
from matplotlib import pyplot as plt

from gluoncv.data import VOCDetection
class VOCLike(VOCDetection):
    CLASSES = ['person', 'dog']
    def __init__(self, root, splits, transform=None, index_map=None, preload_label=True):
        super(VOCLike, self).__init__(root, splits, transform, index_map, preload_label)

train_dataset = VOCLike(root='VOCtemplate', splits=((2018, 'train'),))
print('length of dataset:', len(train_dataset))
print('label example:')
print(train_dataset[0][1])

train_image, train_label = train_dataset[0]
print('Image size (height, width, RGB):', train_image.shape)

bounding_boxes = train_label[:, :4]
print('Num of objects:', bounding_boxes.shape[0])
print('Bounding boxes (num_boxes, x_min, y_min, x_max, y_max):\n',bounding_boxes)
     
class_ids = train_label[:, 4:5]
print('Class IDs (num_boxes, ):\n', class_ids)

ax = utils.viz.plot_bbox(train_image.asnumpy(), bounding_boxes, scores=None,
                   labels=class_ids, class_names=train_dataset.classes)

# see how it looks by rendering the boxes into image
plt.show()



# 读取xml文件内容
# with open('VOCtemplate/VOC2018/Annotations/000001.xml', 'r') as fid:
#     print(fid.read())

2. 数据准备

2.1 voc数据集结构

image.png

在VOCtemplate和.py文件放在相同目录下,

image.png

2.2 Annotations下存放000001.xml格式文件,

<annotation>
    <filename>000001.jpg</filename>
    <size>
        <width>353</width>
        <height>500</height>
        <depth>3</depth>
    </size>
    <object>
        <name>dog</name>
        <difficult>0</difficult>
        <bndbox>
            <xmin>48</xmin>
            <ymin>240</ymin>
            <xmax>195</xmax>
            <ymax>371</ymax>
        </bndbox>
    </object>
    <object>
        <name>person</name>
        <difficult>0</difficult>
        <bndbox>
            <xmin>8</xmin>
            <ymin>12</ymin>
            <xmax>352</xmax>
            <ymax>498</ymax>
        </bndbox>
    </object>
</annotation>

2.3 JPEGImages文件夹存放000001.jpg文件,

000001.jpg

2.4 ImageSets文件夹下存放Main文件夹,Main文件夹下存放train.txt文件.

image.png
image.png
train.txt文件内容:是图片名称,不加后缀.
image.png

3.运行结果:

image.png
image.png

参考

1. https://gluon-cv.mxnet.io/build/examples_datasets/detection_custom.html#sphx-glr-download-build-examples-datasets-detection-custom-py
2. https://blog.csdn.net/zhaotun123/article/details/99676067