可视化gt

 VOC数据格式可视化

import numpy as np
import xml.etree.ElementTree as ET
import os
import cv2
import matplotlib.pyplot as plt 
#import sys
#reload(sys)
#sys.setdefaultencoding("utf-8") 

def GetFileFromThisRootDir(dir,ext = None):
  allfiles = []
  needExtFilter = (ext != None)
  for root,dirs,files in os.walk(dir):   #files是文件夹下所有文件的名称
    for filespath in files:            #依次取文件名
      filepath = os.path.join(root, filespath)    #构成绝对路径
      extension = os.path.splitext(filepath)[1][1:]  #os.path.splitext(path)  #分割路径，返回路径名和文件后缀 其中[1]为后缀.png，再取[1:]得到png
      if needExtFilter and extension in ext:
        allfiles.append(filepath)
      elif not needExtFilter:
        allfiles.append(filepath)
  return allfiles   #返回dir中所有文件的绝对路径

def custombasename(fullname):
    #os.path.basename返回路径的最后一个部分，也就是文件名（os.path.splitext(fullname)[0]是除掉后缀的路径）
    return os.path.basename(os.path.splitext(fullname)[0])  


def load_pascal_annotation(xmlpath):  
    nameboxdict = {}
    gtAttribute = []  #用于进行gt的属性统计
    for index, xmlname in enumerate(xmlpath):
        dets = []
        picname = custombasename(xmlname)
        tree = ET.parse(xmlname)
        objs = tree.findall('object')
        # Load object bounding boxes into a data frame.
        for ix, obj in enumerate(objs):  # 对于该图片上每一个object
            det = []
            name = obj.find('name')    # 找到XML文件中给用于存放gt坐标的<name>模块
            #gt_classes.append(name.text)
            bbox = obj.find('bndbox')    # 找到XML文件中给用于存放gt坐标的<bndbox>模块
            x1 = float(bbox.find('xmin').text)       #左上角列
            y1 = float(bbox.find('ymin').text)       #左上角行
            x2 = float(bbox.find('xmax').text)       #右下角列
            y2 = float(bbox.find('ymax').text)       #右下角行            
            width = x2 - x1 #宽
            if width == 0:
                raise IOError(('width = 0,图片路径为{:s}').format(xmlpath))
                #print('width = 0,图片路径为%s'%xmlpath)
                continue
            else:         
                height = y2 - y1   #高
                area = float(width * height)  #面积
                scale = float(height/width)   #高比宽
                gtdet = [name.text, width, height, area, scale]
                gtAttribute.append(gtdet)
                det = [name.text, x1, y1, x2, y2]
                dets.append(det)
        if (picname not in nameboxdict):
            nameboxdict[picname] = []
            nameboxdict[picname].append(dets)  

    return nameboxdict,gtAttribute

def write_txt(nameboxdict,txtdir): 
    '''
    将读取的gt写入txt
    '''
    if not os.path.exists(txtdir):
        os.makedirs(txtdir)
    for key in nameboxdict:  #key是每一个xlm的名字
        dets = nameboxdict[key][0]   #取相应名字的dets(dets包括该xml所有ob和box)
        gt_txtdir = os.path.join(txtdir, key +'.txt')  #生成相应名字的txt文件，用以写gt
        with open(gt_txtdir, 'wt') as f_out:
            for i in range(len(dets)):
                det = dets[i]
                f_out.write('{:s},{:.1f},{:.1f},{:.1f},{:.1f}\n'.format(det[0],
                                        det[1] + 1, det[2] + 1,
                                        det[3] + 1, det[4] + 1))

def GtVisualise(nameboxdict, JPEGpath, savepath):
    '''
    可视化gt
    '''
    if not os.path.exists(savepath):
        os.makedirs(savepath)
    font = cv2.FONT_HERSHEY_SIMPLEX
    for idx,key in enumerate(nameboxdict):  #key是每一个xlm的名字
        dets = nameboxdict[key][0]   #取相应名字的dets(dets包括该xml所有ob和box)
        imagepath = os.path.join(JPEGpath, key+'.jpg')
        img = cv2.imread(imagepath)      
        for i in range(len(dets)):  
            det = dets[i]
            obname = str(det[0])  #类别名
            left_up_col = int(np.round(det[1]))   #先列后行
            left_up_row = int(np.round(det[2]))
            right_bottom_col = int(np.round(det[3]))
            right_bottom_row = int(np.round(det[4]))    
            cv2.rectangle(img, (left_up_col, left_up_row), (right_bottom_col,right_bottom_row ), (0, 0, 255), thickness=1)
            cv2.rectangle(img, (left_up_col, left_up_row-15), (left_up_col+65 ,left_up_row-2), (255, 0, 0), thickness=-1) #蓝色填充
            cv2.putText(img, obname, (left_up_col, left_up_row-2), font, 0.5,(255, 255, 255), 1) #0.5是大小
        cv2.imwrite(os.path.join(savepath, key + '.jpg'),img)
        #print ('已完成第%d张'%(idx+1))
    return (idx+1)  #返回可视化的图片个数

def Attribute_statistics(gtAttribute):
    gtAttribute = np.array(gtAttribute)
    Object = gtAttribute[:,0]
    Objectclass = np.unique(Object)
    for j in range(len(Objectclass)):
        Objectnum = (Object == Objectclass[j])
        ObjectAttribute = gtAttribute[Objectnum]
        Width = ObjectAttribute[:,1].astype(np.float32)
        Width.sort()
        Height = ObjectAttribute[:,2].astype(np.float32)
        Height.sort()
        Area = ObjectAttribute[:,3].astype(np.float32)
        Area.sort()
        Scale = ObjectAttribute[:,4].astype(np.float32)
        Scale.sort()

        fig = plt.figure()  

        ax1 = fig.add_subplot(221)  
        ax1.hist(Width,20)  
        ax1.set_title('Width')
  
        ax2 = fig.add_subplot(222)  
        ax2.hist(Height,20) 
        ax2.set_title('Height')
  
        ax3 = fig.add_subplot(223)  
        ax3.hist(Area,20) 
        ax3.set_title('Area')
  
        ax4 = fig.add_subplot(224)  
        ax4.hist(Scale,20) 
        ax4.set_title('Scale')
        fig.suptitle(Objectclass[j]) #大图
        plt.show()
                
                 
if __name__ == '__main__':
     xmldir = 'H:\\data\\sewage\\xml'   
     JPEGpath = 'H:\\data\\sewage\\images'   
     gtSavepath = 'H:\\faster_rcnn_for_hangtian\\data\\datasets\\hbbshow'   
     xmlpath = GetFileFromThisRootDir(xmldir)
     nameboxdict, gtAttribute = load_pascal_annotation(xmlpath)
     #Attribute_statistics(gtAttribute)  #统计gt的属性
     num = GtVisualise(nameboxdict, JPEGpath, gtSavepath)

可视化结果：

DOTA数据格式可视化

import cv2 
import os
import numpy as np


thr=0.95
def custombasename(fullname):  
    return os.path.basename(os.path.splitext(fullname)[0])  
  
def GetFileFromThisRootDir(dir,ext = None):  
  allfiles = []  
  needExtFilter = (ext != None)  
  for root,dirs,files in os.walk(dir):  
    for filespath in files:  
      filepath = os.path.join(root, filespath)  
      extension = os.path.splitext(filepath)[1][1:]  
      if needExtFilter and extension in ext:  
        allfiles.append(filepath)  
      elif not needExtFilter:  
        allfiles.append(filepath)  
  return allfiles  

def visualise_gt(label_path, pic_path, newpic_path):
    results =  GetFileFromThisRootDir(label_path)
    for result in results:
        f = open(result,'r')
        lines = f.readlines()
        boxes = []
        for line in lines:
            #score = float(line.strip().split(' ')[8])
            name = result.split('/')[-1]
            box=line.strip().split(' ')[0:8]
            box = np.array(box,dtype = np.float)
            #if float(score)>thr:
            boxes.append(box)
        boxes = np.array(boxes,np.float)
        f.close()   
        filepath=os.path.join(pic_path, name.split('.')[0]+'.jpg')
        im=cv2.imread(filepath)
        #print line3
        for i in range(boxes.shape[0]):
            box =np.array( [[boxes[i][0],boxes[i][1]],[boxes[i][2],boxes[i][3]], \
                            [boxes[i][4],boxes[i][5]],[boxes[i][6],boxes[i][7]]],np.int32)
            box = box.reshape((-1,1,2))
            cv2.polylines(im,[box],True,(0,255,255),2)
        cv2.imwrite(os.path.join(newpic_path,result.split('/')[-1].split('.')[0]+'.jpg'),im)
        #下面是有score的        
        #        x,y,w,h,score=box.split('_')#
        #        score=float(score)
        #        cv2.rectangle(im,(int(x),int(y)),(int(x)+int(w),int(y)+int(h)),(0,0,255),1)
        #        cv2.putText(im,'%3f'%score, (int(x)+int(w),int(y)+int(h)+5),cv2.FONT_HERSHEY_SIMPLEX,0.5,(0,255,0),1)
        #        cv2.imwrite(newpic_path+filename,im)

if __name__ == '__main__':
    root='/home/yantianwang/clone/Hangtian/Orgdata/aug'
    pic_path = os.path.join(root, 'new_JPEGImages') #样本图片路径
    label_path = os.path.join(root, 'new_labelTxt') #DOTA标签的所在路径    
    newpic_path=os.path.join(root,'hbbshow')  
    visualise_gt(label_path, pic_path, newpic_path)