返回至系列文章導(dǎo)航博客

完整項(xiàng)目下載：下載鏈接
【閑魚】https://m.tb.cn/h.52C8psW?tk=fMpwdwfqjz3 CZ3457 「我在閑魚發(fā)布了【舌象數(shù)據(jù)集，詳情見csdn！http://t.csdn.cn】」
點(diǎn)擊鏈接直接打開

1 簡(jiǎn)介

舌體分割是舌診檢測(cè)的基礎(chǔ)，唯有做到準(zhǔn)確分割舌體才能保證后續(xù)訓(xùn)練以及預(yù)測(cè)的準(zhǔn)確性。此部分真正的任務(wù)是在用戶上傳的圖像中準(zhǔn)確尋找到屬于舌頭的像素點(diǎn)。舌體分割屬于生物醫(yī)學(xué)圖像分割領(lǐng)域。分割效果如下：

2 數(shù)據(jù)集介紹

舌象數(shù)據(jù)集包含舌象原圖以及分割完成的二元圖，共979*2張，示例圖片如下：

3 模型介紹

U-Net是一個(gè)優(yōu)秀的語(yǔ)義分割模型，在中e診中U-Net共三部分，分別是主干特征提取部分、加強(qiáng)特征提取部分、預(yù)測(cè)部分。利用主干特征提取部分獲得5個(gè)初步有效的特征層，之后通過加強(qiáng)特征提取部分對(duì)上述獲取到的5個(gè)有效特征層進(jìn)行上采樣并進(jìn)行特征融合。最終獲得了一個(gè)結(jié)合所有特征的有效特征層，并利用最終有效特征層對(duì)像素點(diǎn)進(jìn)行預(yù)測(cè)，找到屬于舌體的像素點(diǎn)。具體操作詳情如下圖所示：

進(jìn)行標(biāo)注后利用PyTorch框架構(gòu)建U-Net模型抓取舌象圖像特征，預(yù)測(cè)舌象圖像標(biāo)簽。為對(duì)模型進(jìn)行評(píng)價(jià)，在訓(xùn)練中計(jì)算每次循環(huán)的平均損失率。最終每張圖的損失了約為2%左右。具體的平均損失率變化如下圖：

訓(xùn)練共歷時(shí)4天，共979張標(biāo)記圖像，最終平均預(yù)測(cè)損失率約為2%。模型預(yù)測(cè)，即舌體分割的效果非常理想，在此展示當(dāng)損失率為40%與損失率為2%時(shí)的分割結(jié)果示例，示例如下圖所示：
（1）損失率為40%時(shí)分割結(jié)果圖

（2）損失率為2%時(shí)分割結(jié)果圖

根據(jù)模型預(yù)測(cè)結(jié)果對(duì)屬于舌體的像素點(diǎn)進(jìn)行匹配提取，將不屬于舌體的部分以墨綠色進(jìn)行填充，最終的舌體分割效果圖如下：

4 代碼實(shí)現(xiàn)細(xì)節(jié)

4.1 相關(guān)文件介紹

notedata文件夾中有分割標(biāo)注圖片、ordata文件夾中有原始圖片、params文件夾中有訓(xùn)練模型文件、result文件夾中有測(cè)試樣例圖片、train_image文件夾中有訓(xùn)練過程圖片。

4.2 utils.py

工具類：由于數(shù)據(jù)集中各個(gè)圖片的大小是不一樣的，為了保障后續(xù)工作可以順利進(jìn)行，這里應(yīng)該定義一個(gè)工具類將圖片可以等比例縮放至256*256（可以改看自己需求）。

from PIL import Image

def keep_image_size_open(path, size=(256, 256)):
    img = Image.open(path)
    temp = max(img.size)
    mask = Image.new('RGB', (temp, temp), (0,0,0))
    mask.paste(img, (0,0))
    mask = mask.resize(size)
    return mask

4.3 data.py

這里主要是將數(shù)據(jù)集中標(biāo)簽圖片與原圖進(jìn)行匹配合并~具體步驟代碼注釋中有詳解！

import os
from torch.utils.data import Dataset
from utils import *
from torchvision import transforms
transform = transforms.Compose([
    transforms.ToTensor()
    ])

class MyDataset(Dataset):
    def __init__(self, path):   #拿到標(biāo)簽文件夾中圖片的名字
        self.path = path
        self.name = os.listdir(os.path.join(path, 'notedata'))
        
    def __len__(self):          #計(jì)算標(biāo)簽文件中文件名的數(shù)量
        return len(self.name)
    
    def __getitem__(self, index):   #將標(biāo)簽文件夾中的文件名在原圖文件夾中進(jìn)行匹配（由于標(biāo)簽是png的格式而原圖是jpg所以需要進(jìn)行一個(gè)轉(zhuǎn)化）
        segment_name = self.name[index] #XX.png
        segment_path = os.path.join(self.path, 'notedata', segment_name)
        image_path = os.path.join(self.path, 'ordata', segment_name.replace('png', 'jpg')) #png與jpg進(jìn)行轉(zhuǎn)化
        
        segment_image = keep_image_size_open(segment_path)  #等比例縮放
        image = keep_image_size_open(image_path)            #等比例縮放
        
        return transform(image), transform(segment_image)

if __name__ == "__main__":
    data = MyDataset("E:/ITEM_TIME/project/UNET/")
    print(data[0][0].shape)
    print(data[0][1].shape)

可見數(shù)據(jù)集已經(jīng)規(guī)整！

4.4 net.py

Unet網(wǎng)絡(luò)的編寫！

from torch import nn
import torch
from torch.nn import functional as F


class Conv_Block(nn.Module):   #卷積
    def __init__(self, in_channel, out_channel):
        super(Conv_Block, self).__init__()
        self.layer = nn.Sequential(
            nn.Conv2d(in_channel, out_channel, 3, 1, 1, padding_mode='reflect', 
                      bias=False),
            nn.BatchNorm2d(out_channel),
            nn.Dropout2d(0.3),
            nn.LeakyReLU(),
            nn.Conv2d(out_channel, out_channel, 3, 1, 1, padding_mode='reflect', 
                      bias=False),
            nn.BatchNorm2d(out_channel),
            nn.Dropout2d(0.3),
            nn.LeakyReLU()
            )
        
    def forward(self, x):
        return self.layer(x)
    
    
class DownSample(nn.Module):    #下采樣
    def __init__(self, channel):
        super(DownSample, self).__init__()
        self.layer = nn.Sequential(
            nn.Conv2d(channel, channel,3,2,1,padding_mode='reflect',
                      bias=False),
            nn.BatchNorm2d(channel),
            nn.LeakyReLU()
            
            )
        
    def forward(self,x):
        return self.layer(x)
    
    
class UpSample(nn.Module):   #上采樣（最鄰近插值法）
    def __init__(self, channel):
        super(UpSample, self).__init__()
        self.layer = nn.Conv2d(channel, channel//2,1,1)
        
    def forward(self,x, feature_map):
        up = F.interpolate(x, scale_factor=2, mode='nearest')
        out = self.layer(up)
        return torch.cat((out,feature_map),dim=1)
    
    
class UNet(nn.Module):
    def __init__(self):
        super(UNet, self).__init__()
        self.c1=Conv_Block(3,64)
        self.d1=DownSample(64)
        self.c2=Conv_Block(64, 128)
        self.d2=DownSample(128)
        self.c3=Conv_Block(128,256)
        self.d3=DownSample(256)
        self.c4=Conv_Block(256,512)
        self.d4=DownSample(512)
        self.c5=Conv_Block(512,1024)
        self.u1=UpSample(1024)
        self.c6=Conv_Block(1024,512)
        self.u2=UpSample(512)
        self.c7=Conv_Block(512,256)
        self.u3=UpSample(256)
        self.c8=Conv_Block(256,128)
        self.u4=UpSample(128)
        self.c9=Conv_Block(128,64)
        
        self.out = nn.Conv2d(64,3,3,1,1)
        self.Th = nn.Sigmoid()

       
        
    def forward(self,x):
        R1 = self.c1(x)
        R2 = self.c2(self.d1(R1))
        R3 = self.c3(self.d2(R2))
        R4 = self.c4(self.d3(R3))
        R5 = self.c5(self.d4(R4))
        
        O1 = self.c6(self.u1(R5,R4))
        O2 = self.c7(self.u2(O1,R3))
        O3 = self.c8(self.u3(O2,R2))
        O4 = self.c9(self.u4(O3,R1))
        
        return self.Th(self.out(O4))
    
if __name__ == "__main__":
    x = torch.randn(2, 3, 256, 256)
    net  = UNet()
    print(net(x).shape)
         

結(jié)果匹配說明沒問題~

4.5 train.py

訓(xùn)練代碼~

from torch import nn
from torch import optim
import torch
from data import *
from net import *
from torchvision.utils import save_image
from torch.utils.data import DataLoader

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
weight_path = 'params/unet.pth'
data_path = 'E:/ITEM_TIME/project/UNET/'
save_path = 'train_image'

if __name__ == "__main__":
    
    dic = []###
    
    data_loader = DataLoader(MyDataset(data_path),batch_size=3,shuffle=True)  #batch_size用3/4都可以看電腦性能
    net = UNet().to(device)
    if os.path.exists(weight_path):
        net.load_state_dict(torch.load(weight_path))
        print('success load weight')
    else:
        print('not success load weight')
        
    opt = optim.Adam(net.parameters())
    loss_fun = nn.BCELoss()
    
    epoch = 1
    while True:
        avg = []###
        for i, (image,segment_image) in enumerate(data_loader):
            image,segment_image = image.to(device),segment_image.to(device)
            
            out_image = net(image)
            train_loss = loss_fun(out_image, segment_image)
            
            opt.zero_grad()
            train_loss.backward()
            opt.step()
            
            if i%5 == 0:
                print('{}-{}-train_loss===>>{}'.format(epoch,i,train_loss.item()))
                
            if i%50 == 0:
                torch.save(net.state_dict(), weight_path)
            #為方便看效果將原圖、標(biāo)簽圖、訓(xùn)練圖進(jìn)行拼接
            _image = image[0]
            _segment_image = segment_image[0]
            _out_image = out_image[0]
            
            img = torch.stack([_image,_segment_image,_out_image],dim=0)
            save_image(img, f'{save_path}/{i}.jpg')
            
            avg.append(float(train_loss.item()))###
            
        
        
        loss_avg = sum(avg)/len(avg)
        
        dic.append(loss_avg)
        
        epoch += 1
    print(dic)
    

可見代碼成功運(yùn)行~上面的損失率是在訓(xùn)練4天后的效果，剛開始肯定很大很差，需要有耐心！

4.6 test.py

測(cè)試代碼，對(duì)圖片進(jìn)行智能分割~

from net import *
from utils import keep_image_size_open
import os
import torch
from data import *
from torchvision.utils import save_image
from PIL import Image
import numpy as np

net = UNet().cpu()  #或者放在cuda上

weights = 'params/unet.pth'  #導(dǎo)入網(wǎng)絡(luò)

if os.path.exists(weights):
    net.load_state_dict(torch.load(weights))
    print('success')
else:
    print('no loading')
    
_input = 'xxxx.jpg'  #導(dǎo)入測(cè)試圖片

img = keep_image_size_open(_input)


img_data = transform(img)
print(img_data.shape)

img_data = torch.unsqueeze(img_data, dim=0)

print(img_data)
out = net(img_data)

save_image(out, 'result/result.jpg')
save_image(img_data, 'result/orininal.jpg')

print(out)

#E:\ITEM_TIME\UNET\ordata\4292.jpg

img_after = Image.open(r"result\result.jpg")
img_before = Image.open(r"result\orininal.jpg")
#img.show()
img_after_array = np.array(img_after)#把圖像轉(zhuǎn)成數(shù)組格式img = np.asarray(image)
img_before_array = np.array(img_before)

shape_after = img_after_array.shape
shape_before = img_before_array.shape

print(shape_after,shape_before)

#將分隔好的圖片進(jìn)行對(duì)應(yīng)像素點(diǎn)還原,即將黑白分隔圖轉(zhuǎn)化為有顏色的提取圖

if shape_after == shape_before:
    height = shape_after[0]
    width = shape_after[1]
    dst = np.zeros((height,width,3))
    for h in range(0,height):
        for w in range (0,width):
            (b1,g1,r1) = img_after_array[h,w]
            (b2,g2,r2) = img_before_array[h,w]
            
            if (b1, g1, r1) <= (90, 90, 90): 
                img_before_array[h, w] = (144,238,144) 
            dst[h,w] = img_before_array[h,w]
    img2 = Image.fromarray(np.uint8(dst))
    img2.save(r"result\blend.png","png")

else:
    print("失?。?)

結(jié)果展示：
（1）原圖（orininal.jpg）：

（2）模型分割圖（result.jpg）：

（3）對(duì)應(yīng)像素點(diǎn)還原圖（blend.png）：就是將（2）中的圖白色的部分用原圖像素點(diǎn)填充，黑色的部分用綠色填充

至此，舌體分割完成！

文章來(lái)源地址http://www.zghlxwxcb.cn/news/detail-419727.html

到了這里，關(guān)于【python-Unet】計(jì)算機(jī)視覺~舌象舌頭圖片分割~機(jī)器學(xué)習(xí)（三）的文章就介紹完了。如果您還想了解更多內(nèi)容，請(qǐng)?jiān)谟疑辖撬阉鱐OY模板網(wǎng)以前的文章或繼續(xù)瀏覽下面的相關(guān)文章，希望大家以后多多支持TOY模板網(wǎng)！