你好! 这是你第一次使用 Markdown编辑器 所展示的欢迎页。如果你想学习如何使用Markdown编辑器, 可以仔细阅读这篇文章,了解一下Markdown的基本语法知识。
在macos windows liunx系统都测试过,本文采用macos
项目代码:https://github.com/qqwweee/keras-yolo3
下载yolo3weights :https://pjreddie.com/darknet/yolo/
将yolo3weights文件夹放到keras-yolo3-master文件夹里
terminal cd 到keras-yolo3-master文件夹
生成现在权重下h5文件:
python3 convert.py yolov3.cfg yolov3.weights model_data/yolo.h5
进行快速测试,看看能不能用。
terminal cd到keras-yolo3-master目录下
python3 yolo_video.py --image
之后让你输入图片路径:(若将图片放在keras-yolo3-master文件夹下,直接输入相对地址即可)
下载VOC2007数据集
下载地址: https://pjreddie.com/projects/pascal-voc-dataset-mirror/
这里面用到的文件夹是Annotation、ImageSets和JPEGImages
其中文件夹Annotation中主要存放xml文件,每一个xml对应一张图像在这里插入图片描述;而ImageSets我们只需要用到Main文件夹,这里面存放的是一些文本文件,通常为train.txt、test.txt等,该文本文件里面的内容是需要用来训练或测试的图像的名字;JPEGImages文件夹中放我们已按统一规则命名好的原始图像。
将自己数据转移到对应目录
// 将自己原始图片,标注过的图片放到VOC数据集相应位置,并生成训练集测试集验证集
//生成训练集测试集验证集对应txt文件,放入相应位置
#%%import os
import random
import shutil #拷贝文件并移动的库path = '/code/kaggle/wechat/' #自己的数据路径img = os.listdir(path + 'pyq') #所有原始图像img_xml = os.listdir(path + 'labeled') #所有xml文件
print('img_num: ',len(img))
print('img_xml_num: ',len(img_xml))#清空VOC数据集文件夹内容
path_img_ori = '/code/kaggle/keras-yolo3-master/VOCdevkit/VOC2007/JPEGImages'
path_xml = '/code/kaggle/keras-yolo3-master/VOCdevkit/VOC2007/Annotations'
img_ = os.listdir(path_img_ori)
xml_ = os.listdir(path_xml)
for img__ in img_:os.remove(os.path.join(path_img_ori,img__))
for xml__ in xml_:os.remove(os.path.join(path_xml,xml__))#生成VOC数据集文件夹内容
k = 0
for i in range(len(img)):path_img_ori = '/code/kaggle/keras-yolo3-master/VOCdevkit/VOC2007/JPEGImages/'path_xml = '/code/kaggle/keras-yolo3-master/VOCdevkit/VOC2007/Annotations/'#拷贝转移文件,并按0,1,2命名新文件#shutil.copyfile(old——path, new-path)shutil.copyfile(path + 'pyq/' + img[i] , path_img_ori + str(k) + '.' + 'jpg')shutil.copyfile(path + 'labeled/' + img_xml[i] , path_xml + str(k) + '.' + img_xml[i].split('.')[-1]) context.append(str(k))k = k+1trainval_percent = 0.2
train_percent = 0.8 #自己定比例
xmlfilepath = path_xml
txtsavepath = '/code/kaggle/keras-yolo3-master/VOCdevkit/VOC2007/ImageSets/Main/'
total_xml = os.listdir(xmlfilepath)num = len(total_xml)
list = range(num)
tv = int(num * trainval_percent)
tr = int(tv * train_percent)
trainval = random.sample(list, tv)
train = random.sample(trainval, tr)#到达的文件路径~/VOCdevkit/VOC2007/ImageSets/Main/trainval.txt'ftrainval = open('/code/kaggle/keras-yolo3-master/VOCdevkit/VOC2007/ImageSets/Main/trainval.txt', 'w')
ftest = open('/code/kaggle/keras-yolo3-master/VOCdevkit/VOC2007/ImageSets/Main/test.txt', 'w')
ftrain = open('/code/kaggle/keras-yolo3-master/VOCdevkit/VOC2007/ImageSets/Main/train.txt', 'w')
fval = open('/code/kaggle/keras-yolo3-master/VOCdevkit/VOC2007/ImageSets/Main/val.txt', 'w')for i in list:name = total_xml[i][:-4] + '\n'if i in trainval:ftrainval.write(name)if i in train:ftest.write(name)else:fval.write(name)else:ftrain.write(name)ftrainval.close()
ftrain.close()
fval.close()
ftest.close()#%%
打开keras-yolo3-master文件夹下voc_annatation.py文件进行修改
import xml.etree.ElementTree as ET
from os import getcwdsets=[('2007', 'train'), ('2007', 'val'), ('2007', 'test')] classes = ["dz_tx"] #这里改为自己标注数据集中的标签名def convert_annotation(year, image_id, list_file):in_file = open('/code/keras-yolo3-master/VOCdevkit/VOC%s/Annotations/%s.xml'%(year, image_id),'rb')tree=ET.parse(in_file)root = tree.getroot()for obj in root.iter('object'):difficult = obj.find('difficult').textcls = obj.find('name').textif cls not in classes or int(difficult)==1:continuecls_id = classes.index(cls)xmlbox = obj.find('bndbox')b = (int(xmlbox.find('xmin').text), int(xmlbox.find('ymin').text), int(xmlbox.find('xmax').text), int(xmlbox.find('ymax').text))list_file.write(" " + ",".join([str(a) for a in b]) + ',' + str(cls_id))wd = getcwd()for year, image_set in sets:image_ids = open('/code/keras-yolo3-master/VOCdevkit/VOC%s/ImageSets/Main/%s.txt'%(year, image_set)).read().strip().split()list_file = open('%s.txt'%( image_set), 'w')for image_id in image_ids:list_file.write('%s/VOCdevkit/VOC%s/JPEGImages/%s.jpg'%(wd, year, image_id))convert_annotation(year, image_id, list_file)list_file.write('\n')list_file.close()
一个标签占一行
直接复制替换原来train.py即可
此时自己在keras-yolo3-master下新建文件夹logs logs下再建文件夹000
run时可能会报错
AttributeError: module ‘keras.backend’ has no attribute ‘control_flow_ops’
解决办法:https://blog.csdn.net/CAU_Ayao/article/details/89312354
可能Tensorboard报错
我发现Tensorboard这行代码是灰色的,所以我把它作为释义不用了
建议先将epoch调小一些进行测试
"""
Retrain the YOLO model for your own dataset.
"""
import numpy as np
import keras.backend as K
from keras.layers import Input, Lambda
from keras.models import Model
from keras.callbacks import TensorBoard, ModelCheckpoint, EarlyStoppingfrom yolo3.model import preprocess_true_boxes, yolo_body, tiny_yolo_body, yolo_loss
from yolo3.utils import get_random_datadef _main():annotation_path = 'train.txt'log_dir = 'logs/000/'classes_path = 'model_data/voc_classes.txt'anchors_path = 'model_data/yolo_anchors.txt'class_names = get_classes(classes_path)anchors = get_anchors(anchors_path)input_shape = (416,416) # multiple of 32, hwmodel = create_model(input_shape, anchors, len(class_names) )train(model, annotation_path, input_shape, anchors, len(class_names), log_dir=log_dir)def train(model, annotation_path, input_shape, anchors, num_classes, log_dir='logs/'):model.compile(optimizer='adam', loss={'yolo_loss': lambda y_true, y_pred: y_pred})#################这个位置########################logging = TensorBoard(log_dir=log_dir)checkpoint = ModelCheckpoint(log_dir + "ep{epoch:03d}-loss{loss:.3f}-val_loss{val_loss:.3f}.h5",monitor='val_loss', save_weights_only=True, save_best_only=True, period=1)batch_size = 10val_split = 0.1with open(annotation_path) as f:lines = f.readlines()np.random.shuffle(lines)num_val = int(len(lines)*val_split)num_train = len(lines) - num_valprint('Train on {} samples, val on {} samples, with batch size {}.'.format(num_train, num_val, batch_size))model.fit_generator(data_generator_wrap(lines[:num_train], batch_size, input_shape, anchors, num_classes),steps_per_epoch=max(1, num_train//batch_size),validation_data=data_generator_wrap(lines[num_train:], batch_size, input_shape, anchors, num_classes),validation_steps=max(1, num_val//batch_size),epochs=500,initial_epoch=0)model.save_weights(log_dir + 'trained_weights.h5')def get_classes(classes_path):with open(classes_path) as f:class_names = f.readlines()class_names = [c.strip() for c in class_names]return class_namesdef get_anchors(anchors_path):with open(anchors_path) as f:anchors = f.readline()anchors = [float(x) for x in anchors.split(',')]return np.array(anchors).reshape(-1, 2)def create_model(input_shape, anchors, num_classes, load_pretrained=False, freeze_body=False,weights_path='model_data/yolo_weights.h5'):K.clear_session() # get a new sessionimage_input = Input(shape=(None, None, 3))h, w = input_shapenum_anchors = len(anchors)y_true = [Input(shape=(h//{0:32, 1:16, 2:8}[l], w//{0:32, 1:16, 2:8}[l], \num_anchors//3, num_classes+5)) for l in range(3)]model_body = yolo_body(image_input, num_anchors//3, num_classes)print('Create YOLOv3 model with {} anchors and {} classes.'.format(num_anchors, num_classes))if load_pretrained:model_body.load_weights(weights_path, by_name=True, skip_mismatch=True)print('Load weights {}.'.format(weights_path))if freeze_body:# Do not freeze 3 output layers.num = len(model_body.layers)-7for i in range(num): model_body.layers[i].trainable = Falseprint('Freeze the first {} layers of total {} layers.'.format(num, len(model_body.layers)))model_loss = Lambda(yolo_loss, output_shape=(1,), name='yolo_loss',arguments={'anchors': anchors, 'num_classes': num_classes, 'ignore_thresh': 0.5})([*model_body.output, *y_true])model = Model([model_body.input, *y_true], model_loss)return model
def data_generator(annotation_lines, batch_size, input_shape, anchors, num_classes):n &#61; len(annotation_lines)np.random.shuffle(annotation_lines)i &#61; 0while True:image_data &#61; []box_data &#61; []for b in range(batch_size):i %&#61; nimage, box &#61; get_random_data(annotation_lines[i], input_shape, random&#61;True)image_data.append(image)box_data.append(box)i &#43;&#61; 1image_data &#61; np.array(image_data)box_data &#61; np.array(box_data)y_true &#61; preprocess_true_boxes(box_data, input_shape, anchors, num_classes)yield [image_data, *y_true], np.zeros(batch_size)def data_generator_wrap(annotation_lines, batch_size, input_shape, anchors, num_classes):n &#61; len(annotation_lines)if n&#61;&#61;0 or batch_size<&#61;0: return Nonereturn data_generator(annotation_lines, batch_size, input_shape, anchors, num_classes)if __name__ &#61;&#61; &#39;__main__&#39;:_main()