本次的YOLO v3实战是基于DataFountain的一个比赛:智能盘点—钢筋数量AI识别,baseline model就选用上次讲解YOLO v3理论YunYang复现的YOLO v3。本次系列也和正常我们做比赛的流程一样分为两部分,这次也是第一部分将会带大家跑通baseline(比赛的话可能会对比多个,这里仅跑YOLO v3),第二部分将会分析baseline出现的问题结合赛题背景进行改进。
import os
train_file=open('data/train_data_VOC/ImageSets/Main/train.txt','w')
test_file=open('data/test_VOC/ImageSets/Main/test.txt','w')
for _,_,train_files in os.walk('data/train_data_VOC/JPEGImages'):
continuefor _,_,test_files in os.walk('data/test_VOC/JPEGImages'):
continuefor file in train_files:
train_file.write(file.split('.')[0]+'\n')
for file in test_files:
test_file.write(file.split('.')[0]+'\n')
import csv
import os, sys
from glob import glob
from PIL import Image
src_img_dir = r'data/train_data_VOC/JPEGImages'#图片地址
src_txt_dir = r'data/yolo'#生成txt地址
img_lists = glob(src_img_dir + '/*jpg')
img_basenames = []
for item in img_lists:
img_basenames.append(os.path.basename(item))
img_names = []
for item in img_basenames:
temp1, temp2 = os.path.splitext(item)
img_names.append(temp1)
c = []
filename = r'/home/cristianoc/tensorflow-yolov3/data/yolo/train.csv'with open(filename) as f:
reader = csv.reader(f)
head_now = next(reader)
l = []
b = []
for cow in reader:
label = cow[0]
l.append(label)
bbox = cow[1]
b.append(bbox)
label = []
for item in l:
temp1, temp2 = os.path.splitext(item)
label.append(temp1)
for img in img_names:
img_file = src_txt_dir + os.sep + img + '.txt'
fp = open(img_file, 'w')
for i in range(len(label)):
if label[i] == img:
fp.write(str(b[i]))
fp.write('\n')
import csv
import os, sys
from glob import glob
from PIL import Image
src_img_dir = r'data/train_data_VOC/JPEGImages'
src_txt_dir = r'data/yolo'
src_xml_dir = r'data/train_data_VOC/Annotations'
img_lists = glob(src_img_dir + '/*jpg')
img_basenames = []
for item in img_lists:
img_basenames.append(os.path.basename(item))
img_names = []
for item in img_basenames:
temp1, temp2 = os.path.splitext(item)
img_names.append(temp1)
for img in img_names:
im = Image.open((src_img_dir + os.sep + img + '.jpg'))
width, height = im.size
gt = open(src_txt_dir + os.sep + img + '.txt').read().splitlines()
xml_file = open((src_xml_dir + os.sep + img + '.xml'), 'w')
xml_file.write('<annotation>\n')
xml_file.write(' <folder>VOC2007</folder>\n')
xml_file.write(' <filename>' + str(img) + '.jpg' + '</filename>\n')
xml_file.write(' <size>\n')
xml_file.write(' <width>' + str(width) + '</width>\n')
xml_file.write(' <height>' + str(height) + '</height>\n')
xml_file.write(' <depth>3</depth>\n')
xml_file.write(' </size>\n')
for img_each_label in gt:
spt = img_each_label.split(' ')
xml_file.write(' <object>\n')
xml_file.write(' <name>' + str('rebar') + '</name>\n')
xml_file.write(' <pose>Unspecified</pose>\n')
xml_file.write(' <truncated>0</truncated>\n')
xml_file.write(' <difficult>0</difficult>\n')
xml_file.write(' <bndbox>\n')
xml_file.write(' <xmin>' + str(spt[0]) + '</xmin>\n')
xml_file.write(' <ymin>' + str(spt[1]) + '</ymin>\n')
xml_file.write(' <xmax>' + str(spt[2]) + '</xmax>\n')
xml_file.write(' <ymax>' + str(spt[3]) + '</ymax>\n')
xml_file.write(' </bndbox>\n')
xml_file.write(' </object>\n')
xml_file.write('</annotation>')
python scripts/voc_annotation.py --data_path data/test_VOC
分别生成我们的训练标注文件和验证标注文件,这样我们的数据就准备好了。config.py
中修改我们读入的类别路径:__C.YOLO.CLASSES = "./data/classes/class.names"
config.py
下修改对应训练标注文件和验证标注文件的路径,改为刚才生成好的即可。__C.YOLO.ORIGINAL_WEIGHT
是convert_weights.py
转换权重文件的源文件,__C.YOLO.DEMO_WEIGHT
是转换后生成的目标权重文件(用于将在COCO预训练好的权重文件转换后生成预训练模型)__C.YOLO.DEMO_WEIGHT
就是预训练模型。python convert_weight.py --train_from_coco
开始转换: