【BUG】VOC格式数据集报错

问题描述

在运行tools/x2coco.py 时，一直报错
assert xmax > xmin and ymax > ymin, “Box size error.”
AssertionError: Box size error

解决方法

修改x2coco.py，查看是哪里出了问题。

def voc_get_coco_annotation(obj, label2id):
    label = obj.findtext('name')
    filename = obj.findtext('filename') #加上名字
    assert label in label2id, "label is not in label2id."
    category_id = label2id[label]
    bndbox = obj.find('bndbox')
    xmin = int(bndbox.findtext('xmin'))
    ymin = int(bndbox.findtext('ymin'))
    xmax = int(bndbox.findtext('xmax'))
    ymax = int(bndbox.findtext('ymax'))
    #修改这里
    assert xmax > xmin and ymax > ymin, "{}Box size error.{}and{}".format(filename,xmax,xmin)
    o_width = xmax - xmin
    o_height = ymax - ymin
    anno = {
        'area': o_width * o_height,
        'iscrowd': 0,
        'bbox': [xmin, ymin, o_width, o_height],
        'category_id': category_id,
        'ignore': 0,
    }
    return anno

最后发现是有一个文件标注有问题，xmax和xmin一样，用labelImg打开发现有一个错误标注，删掉后再运行就没有报错。
附上一个检测数据标注的代码

# chechk if xmin coordinates have 0 value,todo:checkif max is greater than width

import numpy as np
from PIL import Image
import xml.dom.minidom
import os

if __name__ == '__main__':
    badFilesList = []
    curDir = '/home/work/train/JPEGImages'
   
    path = "/home/aistudio/work/train/Annotations/"
    Assertion_provoker_y = []
    Assertion_provoker_x = []
    files = os.listdir(path)
    for file in files:
        file = os.path.join(path, file)
        dom = xml.dom.minidom.parse(file)
        root = dom.documentElement
        size = dom.getElementsByTagName("size")

        height = dom.getElementsByTagName("height")[0]
        width = dom.getElementsByTagName("width")[0]
        width = width.childNodes[0].data
        height = height.childNodes[0].data

        objects = dom.getElementsByTagName("object")
        for object in objects:
            bndbox = object.getElementsByTagName('bndbox')[0]  ########
            xmin = bndbox.getElementsByTagName('xmin')[0]
            xmin_data = xmin.childNodes[0].data
            ymin = bndbox.getElementsByTagName('ymin')[0]
            ymin_data = ymin.childNodes[0].data
            xmax = bndbox.getElementsByTagName('xmax')[0]
            xmax_data = xmax.childNodes[0].data
            ymax = bndbox.getElementsByTagName('ymax')[0]
            ymax_data = ymax.childNodes[0].data
            # check if the xmin or yin is 0
            if (int(xmin_data) == 0):
                print(file, "the x of it is 0")
            if (int(ymin_data) == 0):
                print(file, "the y of it is 0")
            ##check if min is greater than max
            if (int(ymin_data) >  int(ymax_data)) or ymax_data ==ymin_data:
                Assertion_provoker_y.append(file)
            if (int(xmin_data) > int(xmax_data)) or xmax_data ==xmin_data:
                Assertion_provoker_x.append(file)
            # check if x is greater than width,or y is greater than height
            if (int(xmax_data) > int(width)) :
                Assertion_provoker_x.append(file)
            if (int(xmin_data) > int(width)):
                Assertion_provoker_x.append(file)
            if (int(ymax_data) > int(height)):
                Assertion_provoker_y.append(file)
            if (int(ymin_data) > int(height)):
                Assertion_provoker_y.append(file)
    print("checking assertion error done,already checked", len(files),
          "files,with {} y_provoker".format(len(Assertion_provoker_y)))
    print("checking assertion error done,already checked", len(files),
          "files,with {} x_provoker".format(len(Assertion_provoker_x)))
    print(Assertion_provoker_x)
#######################################################