问题遇到的现象和发生背景
建立了一个二分类的CNN网络，但是没有识别能力。一个类是西方龙图片（很多相似），一个类是其他图片（差别很大）.为什么会没有识别能力。很迷惑
问题相关代码，请勿粘贴截图
import torch
import torchvision
import os
from PIL import Image
from torch.utils.data import Dataset
import torch.utils.data as Data
import torch.nn as nn
import torch.nn.functional as F
import numpy as np
class MyData(Dataset):
def __init__(self,root,labal):
self.dic={'dragon':[1,0],'else':[0,1]}
self.root=root
self.labal=labal
self.path=os.path.join(self.root,self.labal)
self.imgpass=os.listdir(self.path)
def __getitem__(self, item):
imgname=self.imgpass[item]
imgitempass=os.path.join(self.root,self.labal,imgname)
img =Image.open(imgitempass).convert('RGB')
labal=self.dic[self.labal]
trains = torchvision.transforms.Resize([256, 256])
trains1 = torchvision.transforms.ToTensor()
img1=trains(img)
return trains1(img1),torch.tensor(labal,dtype=torch.float32)
def __len__(self):
return len(self.imgpass)
mydata=MyData(r'C:\Users\xiaob\Desktop\西方龙',r'dragon')
batch = 3
loader = Data.DataLoader(
dataset=mydata,
batch_size=batch,
drop_last=True,
num_workers=0
)
class Model(nn.Module):
def __init__(self):
super(Model, self).__init__()
self.conv1=nn.Conv2d(3,32,5,padding=2,stride=1)
self.maxpoo1=nn.MaxPool2d(2)
self.conv2=nn.Conv2d(32,32,5,padding=2,stride=1)
self.maxpoo2 = nn.MaxPool2d(2)
self.conv3 = nn.Conv2d(32, 64, 5, padding=2, stride=1)
self.maxpoo3 = nn.MaxPool2d(2)
self.conv4 = nn.Conv2d(64, 64, 5, padding=2, stride=1)
self.maxpoo4 = nn.MaxPool2d(2)
self.maxpoo5 = nn.MaxPool2d(2)
self.flatten = nn.Flatten()
self.Linear1 = nn.Linear(64*64, 64)
self.Linear2 = nn.Linear(64, 16)
self.Linear3 = nn.Linear(16, 2)
def forward(self,x):
x = self.conv1(x)
x = nn.functional.relu(x)
x = self.maxpoo1(x)
x = self.conv2(x)
x = nn.functional.relu(x)
x = self.maxpoo2(x)
x = self.conv3(x)
x = nn.functional.relu(x)
x = self.maxpoo3(x)
x = self.conv4(x)
x = nn.functional.relu(x)
x = self.maxpoo4(x)
x = self.maxpoo5(x)
x = x.view(-1,4096)
x = self.Linear1(x)
x = self.Linear2(x)
x = self.Linear3(x)
return x
model=Model()
model = torch.load(r'C:\Users\xiaob\Desktop\西方龙\model\mo.pt')
model.eval()
print('mode is complit ,1 for trainning ,0 for test')
x=input(':')
if (int(x)==1) :
loss = nn.CrossEntropyLoss()
optim=torch.optim.SGD(model.parameters(),lr=0.01)
for epoch in range(100):
runningloss=0.0
for data in loader:
img,targets=data
ouputs=model(img)
result_loss=loss(ouputs,targets)
optim.zero_grad()
result_loss.backward()
optim.step()
runningloss=runningloss+result_loss
print(runningloss)
torch.save(model,r'C:\Users\xiaob\Desktop\西方龙\model\mo.pt')
else:
x2 = input('root:')#测试图片地址
img1=Image.open(x2).convert('RGB')
print('openseccses')
trains = torchvision.transforms.Resize([256, 256])
trains1 = torchvision.transforms.ToTensor()
img1=trains(img1)
img1 = trains1(img1)
print('trainsseccses')
ouputs = model(img1)
print('outsseccses')
kk=torch.nn.functional.softmax(ouputs,dim=1)[0]
kk=kk.detach().numpy().tolist()[0]
print(kk)
运行结果及报错内容
不知道为什么，啥图片放上去都是0.99997的概率.白云图片，狗图片，放上去全是0.999997
我的解答思路和尝试过的方法
我觉得是因为只训练了龙图片，于是我又拿了一个狗类的图片去训练。然后当狗的识别率上升的时候，龙图片的识别率急速下降。反之亦然。我在想是不是激活函数的问题
我想要达到的结果
想知道如何构建一个识别西方龙图片的网络，以及我做错了什么。本人qq248167069
