由于飞桨直接提供了算力,我在本地只配置了cpu版本的
本次使用内置的Resnet50来做猫狗分类
大致流程如下
数据处理
基础配置
CONFIG = {
'model_save_dir': "./output/zodiac", # 模型保存位置
'num_classes': 2, # 结果分类有多少种
'total_images': 7096,
'epochs': 20, # 训练多少轮
'batch_size': 32, # 每次取多少样本训练
'image_shape': [3, 224, 224], # 图片大小
'LEARNING_RATE': { # 学习率
'params': {
'lr': 0.00375
}
},
'OPTIMIZER': {
'params': {
'momentum': 0.9
},
'regularizer': {
'function': 'L2',
'factor': 0.000001
}
},
'LABEL_MAP': [ # 分类标签
"cat",
"dog",
]
}
def get(full_path):
for id, name in enumerate(full_path.split('.')):
if id == 0:
config = CONFIG
config = config[name]
return config
生成标注数据
import io
import os
from PIL import Image
# 数据集根目录
DATA_ROOT = r'/home/aistudio/data/data161626/train'
# 标签List
LABEL_MAP = get('LABEL_MAP')
# 标注生成函数
def generate_annotation(mode):
# 建立标注文件
with open('{}/{}.txt'.format(DATA_ROOT, mode), 'w') as f:
# 对应每个用途的数据文件夹,train/valid/test
train_dir = '{}/{}'.format(DATA_ROOT, mode)
# 遍历文件夹,获取里面的分类文件夹
for path in os.listdir(train_dir):
# 标签对应的数字索引,实际标注的时候直接使用数字索引
label_index = LABEL_MAP.index(path)
# 图像样本所在的路径
image_path = '{}/{}'.format(train_dir, path)
# 遍历所有图像
for image in os.listdir(image_path):
# 图像完整路径和名称
image_file = '{}/{}'.format(image_path, image)
try:
# 验证图片格式是否ok
with open(image_file, 'rb') as f_img:
image = Image.open(io.BytesIO(f_img.read()))
image.load()
if image.mode == 'RGB':
f.write('{}\t{}\n'.format(image_file, label_index))
except:
continue
generate_annotation('train') # 生成训练集标注文件
generate_annotation('valid') # 生成验证集标注文件
generate_annotation('test') # 生成测试集标注文件
自定义数据集
import paddle
import paddle.vision.transforms as T
import numpy as np
from PIL import Image
__all__ = ['ZodiacDataset']
# 定义图像的大小
image_shape = get('image_shape') # 'image_shape': [3, 224, 224],
IMAGE_SIZE = (image_shape[1], image_shape[2])
class ZodiacDataset(paddle.io.Dataset):
"""
猫狗数据集类的定义
"""
def __init__(self, mode='train'):
"""
初始化函数
"""
assert mode in ['train', 'test', 'valid'], 'mode is one of train, test, valid.' # 判断参数合法性
self.data = []
"""
根据不同模式选择不同的数据标注文件
"""
with open(r'/home/aistudio/data/data161626/train/{}.txt'.format(mode)) as f:
for line in f.readlines():
info = line.strip().split('\t')
if len(info) > 0:
self.data.append([info[0].strip(), info[1].strip()]) # 进行切分形成数组,每个数组包含图像的地址和label
if mode == 'train':
self.transforms = T.Compose([
T.RandomResizedCrop(IMAGE_SIZE), # 随机裁剪大小,裁剪地方不同等于间接增加了数据样本 300*300-224*224
T.RandomHorizontalFlip(0.5), # 随机水平翻转,概率0.5,也是等于得到一个新的图像
T.ToTensor(), # 数据的格式转换和标准化 HWC => CHW
T.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]) # 图像归一化
])
else: # 评估模式:没必要进行水平翻转增加样本量了,主要是想看看效果
self.transforms = T.Compose([
T.Resize(256), # 图像大小修改
T.RandomCrop(IMAGE_SIZE), # 随机裁剪,
T.ToTensor(), # 数据的格式转换和标准化 HWC => CHW
T.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]) # 图像归一化
])
def __getitem__(self, index):
"""
根据索引获取单个样本
"""
image_file, label = self.data[index]
image = Image.open(image_file)
# 转成RGB模式,三通道的
if image.mode != 'RGB':
image = image.convert('RGB')
image = self.transforms(image) # 得到预处理后的结果
return image, np.array(label, dtype='int64') # 对label做个数据转换,int类型转成numpy
def __len__(self):
"""
获取样本总数
"""
return len(self.data)
训练
import paddle
import numpy as np
train_dataset = ZodiacDataset(mode='train')
valid_dataset = ZodiacDataset(mode='valid')
network = paddle.vision.models.resnet50(num_classes=get('num_classes'), pretrained=True)
model = paddle.Model(network)
model.summary((-1, ) + tuple(get('image_shape')))
EPOCHS = get('epochs')
BATCH_SIZE = get('batch_size')
def create_optim(parameters):
step_each_epoch = get('total_images') # get('batch_size')
lr = paddle.optimizer.lr.CosineAnnealingDecay(learning_rate=get('LEARNING_RATE.params.lr'),
T_max=step_each_epoch * EPOCHS)
return paddle.optimizer.Momentum(learning_rate=lr,
parameters=parameters,
weight_decay=paddle.regularizer.L2Decay(get('OPTIMIZER.regularizer.factor'))) #正则化来提升精度
# 模型训练配置
model.prepare(create_optim(network.parameters()), # 优化器
paddle.nn.CrossEntropyLoss(), # 损失函数
paddle.metric.Accuracy(topk=(1, 5))) # 评估指标
# 训练可视化VisualDL工具的回调函数
visualdl = paddle.callbacks.VisualDL(log_dir='visualdl_log')
# 启动模型全流程训练
model.fit(train_dataset, # 训练数据集
valid_dataset, # 评估数据集
epochs=EPOCHS, # 总的训练轮次
batch_size=BATCH_SIZE, # 批次计算的样本量大小
shuffle=True, # 是否打乱样本集
verbose=1, # 日志展示格式
save_dir='./chk_points/', )# 分阶段的训练模型存储路径
# callbacks=[visualdl]) # 回调函数使用
model.save(get('model_save_dir')) # 这里添加train=False参数可将模型保存为静态模型
预测
使用预测样本进行预测
import paddle
predict_dataset = ZodiacDataset(mode='test')
print('测试数据集样本量:{}'.format(len(predict_dataset)))
from paddle.static import InputSpec
# 网络结构示例化
network = paddle.vision.models.resnet50(num_classes=get('num_classes'))
# 模型封装
model_2 = paddle.Model(network, inputs=[InputSpec(shape=[-1] + get('image_shape'), dtype='float32', name='image')])
# 训练好的模型加载
model_2.load(get('model_save_dir'))
print(get('model_save_dir'))
# 模型配置
model_2.prepare()
# 执行预测
result = model_2.predict(predict_dataset)
print(result)
model_2.save("./out/info", training=False)
# 样本映射
LABEL_MAP = get('LABEL_MAP')
# 随机取样本展示
indexs = np.random.randint(1, 646, size=20)
for idx in indexs:
predict_label = np.argmax(result[0][idx])
real_label = predict_dataset[idx][1]
print('样本ID:{}, 真实标签:{}, 预测值:{}'.format(idx, LABEL_MAP[real_label], LABEL_MAP[predict_label]))
单张图片预测验证
import numpy as np
import paddle
import paddle.vision.transforms as T
from PIL import Image
import matplotlib.pyplot as plt
def load_image(image_file):
im = Image.open(image_file)
# 将图片调整为跟训练数据一样的大小 224*224, 设定ANTIALIAS,即抗锯齿.resize是缩放
im = im.resize((224, 224), Image.ANTIALIAS)
# 建立图片矩阵 类型为float32
im = np.array(im).astype(np.float32)
# 矩阵转置
im = im.transpose((2, 0, 1))
# 将像素值从【0-255】转换为【0-1】
im = im / 255.0
# print(im)
im = np.expand_dims(im, axis=0)
# 保持和之前输入image维度一致
return im
def sigle_pic_foreast():
img_file = r"/home/aistudio/data/data161626/train/test/cat/cat_5.jpg"
# img_file = r"C:\Users\Public\Pictures\Sample Pictures\45fdeb26c9bd4af3b129c99424af932a.jpg"
image = load_image(img_file)
from paddle.static import InputSpec
# 网络结构示例化
network = paddle.vision.models.resnet50(num_classes=get('num_classes'))
# 模型封装
model_2 = paddle.Model(network, inputs=[InputSpec(shape=[-1] + get('image_shape'), dtype='float32', name='image')])
# 训练好的模型加载
model_2.load(get('model_save_dir'))
# 模型配置
model_2.prepare()
result = model_2.predict((image,))
print(result)
sigle_pic_foreast()
转换为静态图,并进行单张图片验证
import time
import numpy as np
import paddle
from PIL import Image
from paddle import fluid
from paddle.static import InputSpec
# 网络结构示例化
from config import get
# 转为静态图
def get_model():
network = paddle.vision.models.resnet50(num_classes=get('num_classes'), pretrained=True)
# 模型封装
model_2 = paddle.Model(network, inputs=[InputSpec(shape=[-1] + get('image_shape'), dtype='float32', name='image')])
# 训练好的模型加载
model_2.load("./out/info")
# 模型配置
model_2.prepare()
# 保存为训练模式
model_2.save("./out", training=False)
# 预处理图片
def load_image(file):
img = Image.open(file)
# 统一图像大小
img = img.resize((224, 224), Image.ANTIALIAS)
# 转换成numpy值
img = np.array(img).astype(np.float32)
# 转换成CHW
img = img.transpose((2, 0, 1))
# 转换成BGR
img = img[(2, 1, 0), :, :] / 255.0
img = np.expand_dims(img, axis=0)
return img
# img_file = r"E:\train\train\dog\dog_2001.jpg"
img_file = r"C:\Users\Public\Pictures\Sample Pictures\TBj0nti43n6epO (1).jpg"
# img_file = r"C:\Users\Lenovo\Downloads\969d9e9be2909b3ae2433ac012a1f4c9.jpeg"
# img_file = r"C:\Users\Lenovo\Downloads\bbd7aaacbec520666e8c38457d19d393.jpeg"
def deploy(img_file):
# 开始预测图片
img = load_image(img_file)
loaded_model = paddle.jit.load("./out/info")
result = loaded_model(img)
LABEL_MAP = get('LABEL_MAP')
predict_label = np.argmax(result[0])
ans = LABEL_MAP[predict_label]
print(result, predict_label)
print("预测值:{},预测准确度:{}".format(ans, result[0][0][0][predict_label].numpy()[0]))
if __name__ == '__main__':
t1 = time.time()
deploy(img_file)
print(time.time()-t1)
代码和数据集下载:https://url99.ctfile.com/f/37358799-627781656-ac9444?p=8747 (访问密码: 8747)