2023年11月29日发(作者：)
【猫狗数据集】使⽤top1和top5准确率衡量模型
数据集下载地址：
链接：/s/1l1AnBgkAAEhh0vI5_loWKw
提取码：2xq4
创建数据集：
读取数据集：
进⾏训练：
保存模型并继续进⾏训练：
加载保存的模型并测试：
划分验证集并边训练边验证：
使⽤学习率衰减策略并边训练边测试：
利⽤tensorboard可视化训练和测试过程：
从命令⾏接收参数：
epoch、batchsize、step之间的关系：
之前使⽤的仅仅是top1准确率。在图像分类中，⼀般使⽤top1和top5来衡量分类模型的好坏。下⾯来看看。
⾸先在util下新建⼀个⽂件，向⾥⾯加⼊计算top1和top5准确率的代码：
import torch
def accu(output, target, topk=(1,)):
"""Computes the accuracy over the k top predictions for the specified values of k"""
with _grad():
maxk = max(topk)
batch_size = (0)
_, pred = output.topk(maxk, 1, True, True)
pred = pred.t()
correct = ((1, -1).expand_as(pred))
res = []
for k in topk:
correct_k = correct[:k].view(-1).float().sum(0, keepdim=True)
(correct__(100.0 / batch_size))
return res
重点就是topk()函数：
(input, k, dim=None, largest=True, sorted=True, out=None) -> (Tensor, LongTensor)
input：输⼊张量
k：指定返回的前⼏位的值
dim：排序的维度
largest：返回最⼤值
sorted：返回值是否排序
out：可选输出张量
需要注意的是我们这⾥只有两类，因此不存在top5。因此如果设置参数topk=(1,5)，则会报错：
RuntimeError:invalid argument 5:k not in range for
dimension at /pytorch/ate ...
因此我们只能设置topk=(1,2)，⽽且top2的值肯定是100%。最终res中第⼀位存储的是top1准确率，第⼆位存储的是top2准确率。
然后修改对应的：
import torch
from tqdm import tqdm
from tensorflow import summary
import datetime
from utils import acc
"""
current_time = str(().timestamp())
train_log_dir = '/content/drive/My Drive/colab notebooks/output/tsboardx/train/' + current_time
test_log_dir = '/content/drive/My Drive/colab notebooks/output/tsboardx/test/' + current_time
val_log_dir = '/content/drive/My Drive/colab notebooks/output/tsboardx/val/' + current_time
train_summary_writer = _file_writer(train_log_dir)
val_summary_writer = _file_writer(val_log_dir)
test_summary_writer = _file_writer(test_log_dir)
"""
class Trainer:
def __init__(self,criterion,optimizer,model):
ion=criterion
zer=optimizer
=model
def get_lr(self):
for param_group in _groups:
return param_group['lr']
def loop(self,num_epochs,train_loader,val_loader,test_loader,scheduler=None,acc1=0.0):
1=acc1
for epoch in range(1,num_epochs+1):
lr=_lr()
print("epoch：{}，lr：{:.6f}".format(epoch,lr))
(train_loader,epoch,num_epochs)
(val_loader,epoch,num_epochs)
(test_loader,epoch,num_epochs)
if scheduler is not None:
()
def train(self,dataloader,epoch,num_epochs):
()
with _grad():
self._iteration_train(dataloader,epoch,num_epochs)
def val(self,dataloader,epoch,num_epochs):
()
with _grad():
self._iteration_val(dataloader,epoch,num_epochs)
def test(self,dataloader,epoch,num_epochs):
'train_loss':train_,
'train_acc':train_,
}
save_path="/content/drive/My Drive/colab notebooks/output/"
(state,save_path+"/resnet18_final_v2"+".t7")
"""
t_loss = train_,
t_top1 = train_
t_top2 = train_
return t_loss,t_top1,t_top2
def _iteration_val(self,dataloader,epoch,num_epochs):
#total_step=len(dataloader)
#tot_loss = 0.0
#correct = 0
#for i ,(images, labels) in enumerate(dataloader):
t_top1 = test_
t_top2 = test_
"""
with test_summary__default():
('loss', test_, epoch)
('accuracy', test_, epoch)
"""
"""
if epoch_acc > 1:
state = {
"model": _dict(),
"optimizer": _dict(),
"epoch": epoch,
发布评论