水果识别-训练

水果识别

• 更改voc_label.py生成label文件
• 修改要训练的网络配置文件
• 修改voc.names
• 修改voc.data
• 训练

1. 更改voc_label.py生成label文件，该文件在/darknet/scripts/目录下

   import xml.etree.ElementTree as ET
   import pickle
   import os
   from os import listdir, getcwd
   from os.path import join
   
   # sets=[('2012', 'train'), ('2012', 'val'), ('2007', 'train'), ('2007', 'val'), ('2007', 'test')] 
   # classes = ["aeroplane", "bicycle", "bird", "boat", "bottle", "bus", "car", "cat", "chair", "cow", "diningtable", "dog", "horse", "motorbike", "person", "pottedplant", "sheep", "sofa", "train", "tvmonitor"]
   
   sets=[('2007', 'train')] 这里只用到训练，数据集命名为VOC2017
   classes = ["apple", "pear", "banana"] 只用了三个类别作为展示
   
   
   def convert(size, box):
      dw = 1./(size[0])
      dh = 1./(size[1])
      x = (box[0] + box[1])/2.0 - 1
      y = (box[2] + box[3])/2.0 - 1
      w = box[1] - box[0]
      h = box[3] - box[2]
      x = x*dw
      w = w*dw
      y = y*dh
      h = h*dh
      return (x,y,w,h)
   
   def convert_annotation(year, image_id):
      in_file = open('VOCdevkit/VOC%s/Annotations/%s.xml'%(year, image_id))
      out_file = open('VOCdevkit/VOC%s/labels/%s.txt'%(year, image_id), 'w')
      tree=ET.parse(in_file)
      root = tree.getroot()
      size = root.find('size')
      w = int(size.find('width').text)
      h = int(size.find('height').text)
   
      for obj in root.iter('object'):
          difficult = obj.find('difficult').text
          cls = obj.find('name').text
          if cls not in classes or int(difficult)==1:
              continue
          cls_id = classes.index(cls)
          xmlbox = obj.find('bndbox')
          b = (float(xmlbox.find('xmin').text), float(xmlbox.find('xmax').text), float(xmlbox.find('ymin').text), float(xmlbox.find('ymax').text))
          bb = convert((w,h), b)
          out_file.write(str(cls_id) + " " + " ".join([str(a) for a in bb]) + '\n')
   
   wd = getcwd()
   
   for year, image_set in sets:
      if not os.path.exists('VOCdevkit/VOC%s/labels/'%(year)):
          os.makedirs('VOCdevkit/VOC%s/labels/'%(year))
      image_ids = open('VOCdevkit/VOC%s/ImageSets/Main/%s.txt'%(year, image_set)).read().strip().split()
      list_file = open('%s_%s.txt'%(year, image_set), 'w')
      for image_id in image_ids:
          list_file.write('%s/VOCdevkit/VOC%s/JPEGImages/%s.jpg\n'%(wd, year, image_id))
          convert_annotation(year, image_id)
      list_file.close()
   

   修改好后执行python voc_label.py 在scripts/VOCdevkit/VOC2007目录下生成了labels文件夹，文件夹里面长这个样子的： 同时会在/darknet/scripts目录下生成2007_train.txt文件
2. 配置文件的修改，我用到的yolov3，所以就修改/cfg/yolove-voc.cfg

   [net]
   # Testing
   # batch=1
   # subdivisions=1
   # Training       # 训练的时候把Testing的参数注释掉，把Training的参数取消注释
   batch=64
   subdivisions=16
   width=416
   height=416
   channels=3
   momentum=0.9
   decay=0.0005
   angle=0
   saturation = 1.5
   exposure = 1.5
   hue=.1
   
   learning_rate=0.001          # 根据需要修改学习率，就是梯度下降的速率，越大训练速度越快但会牺牲准确率
   burn_in=1000
   max_batches = 100000     # 训练的总次数
   policy=steps
   steps=100,10000,70000,80000,90000        # 训练到相应次数后学习率变化
   scales=10,10,.1,.1,.1        # 学习率变化情况，是累乘操作
   
   ......
   
   
   [convolutional]
   size=1
   stride=1
   pad=1
   filters=24           # filters = 3 × (calsses + 5)
   # yolov2和tiny-yolov2中为 filter= num ×（classes + coords + 1）
   activation=linear
   
   [yolo]
   mask = 0,1,2
   anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
   classes=3        # 修改类别数量
   num=9
   jitter=.3
   ignore_thresh = .5
   truth_thresh = 1
   random=1         # 为1时会启用Multi-Scale Training，随机使用不同尺寸的图片进行训练，显存小可以置为0
   

3. 修改voc.names，该文件在/darknet/data目录下

   apple
   pear
   banana
   

   该文件下记录所有类别名称。
4. 修改voc.data，该文件在/darknet/cfg目录下

   classes= 3 //类别数量
   train  = /home/embedded/lm/darknet/scripts/2007_train.txt    //指定训练样本
   // valid  = /home/pjreddie/data/voc/2007_test.txt  指定测试数据，训练用不到
   names = /home/embedded/lm/darknet/data/voc.names //指向上一步修改的voc.names文件
   backup = /home/embedded/lm/darknet/backup/yolov3_voc_weights //指定训练后的权值放在哪儿
   

5. 训练不要与训练权重直接训练：

   ./darknet detector train cfg/voc.data cfg/yolov3_voc.cfg
   

   也可以下载预训练权重进行训练：

   ./darknet detector train cfg/voc.data cfg/yolov3_voc.cfg yolo.weights
   

   训练过程中可以随时停止，停止以后可以在保存的权重处接着开始训练，保存的权重就当做预训练权重，yolov3开始没训练100次保存一次权重，过1000以后每训练10000次保存一次权重。然后就等待漫长的训练过程……