#coding=gbk
'''
GPU上面的環(huán)境變化太復(fù)雜，這里我直接給出在筆記本CPU上面的運(yùn)行時間結(jié)果

由于方式3需要將tensor轉(zhuǎn)換到GPU上面，這一過程很消耗時間，大概需要十秒，故而果斷拋棄這樣的做法

img (168, 300, 3)
sub div in numpy,time 0.0110
sub div in torch.tensor,time 0.0070
sub div in torch.tensor with torchvision.transforms,time 0.0050
tensor1=tensor2
tensor2=tensor3


img (1079, 1349, 3)
sub div in numpy,time 0.1899
sub div in torch.tensor,time 0.1469
sub div in torch.tensor with torchvision.transforms,time 0.1109
tensor1=tensor2
tensor2=tensor3


耗時最久的是numpy，其次是轉(zhuǎn)換成torch.tensor，最快的是直接使用torchvision.transforms
我現(xiàn)在在GPU上面跑的程序GPU利用率特別低（大多數(shù)時間維持在2%左右，只有很少數(shù)的時間超過80%）
然后設(shè)置打印點(diǎn)調(diào)試程序時發(fā)現(xiàn)，getitem()輸出一張圖像的時間在0.1秒的數(shù)量級，這對于GPU而言是非常慢的
因?yàn)镚PU計算速度很快，CPU加載圖像和預(yù)處理圖像的速度趕不上GPU的計算速度，就會導(dǎo)致顯卡大量時間處于空閑狀態(tài)
經(jīng)過對于圖像I/O部分代碼的定位，發(fā)現(xiàn)是使用numpy減去圖像均值除以方差這一操作浪費(fèi)了太多時間，而且輸入圖像的分辨率越大，
所消耗的時間就會更多
原則上，圖像預(yù)處理每個階段的時間需要維持在0.01秒的數(shù)量級

所以，

'''

import numpy as np
import time
import torch
import torchvision.transforms as transforms
import cv2
# img_path='/ssddata2/wyx/detection/ead_stage12/stage12_img/WL_00387.jpg'
img_path='F:\\2\\00004.jpg'
PIXEL_MEANS =(0.485, 0.456, 0.406)  #RGB format mean and variances
PIXEL_STDS = (0.229, 0.224, 0.225)

#輸入文件路徑，輸出的應(yīng)該是轉(zhuǎn)換成torch.tensor的標(biāo)準(zhǔn)形式

#方式一  在numpy中進(jìn)行減去均值除以方差，最后轉(zhuǎn)換成torch.tensor
one_start=time.time()
img=cv2.imread(img_path)
img=img[:,:,::-1]
img=img.astype(np.float32, copy=False)
img/=255.0
img-=np.array(PIXEL_MEANS)
img/=np.array(PIXEL_STDS)
tensor1=torch.from_numpy(img.copy())
tensor1=tensor1.permute(2,0,1)
one_end=time.time()
print('sub div in numpy,time {:.4f}'.format(one_end-one_start))

del img

#方式二 轉(zhuǎn)換成torch.tensor，再減去均值除以方差
two_start=time.time()
img=cv2.imread(img_path)
img=img[:,:,::-1]
print('img',img.shape,np.min(img),np.min(img))
tensor2=torch.from_numpy(img.copy()).float()
tensor2/=255.0
tensor2-=torch.tensor(PIXEL_MEANS)
tensor2/=torch.tensor(PIXEL_STDS)
tensor2=tensor2.permute(2,0,1)
two_end=time.time()
print('sub div in torch.tensor,time {:.4f}'.format(two_end-two_start))

del img

#方式三 轉(zhuǎn)換成torch.tensor，再放到GPU上面，最后減去均值除以方差
# three_start=time.time()
# img=cv2.imread(img_path)
# img=img[:,:,::-1]
# tensor3=torch.from_numpy(img.copy()).cuda().float()
# tensor3-=torch.tensor(PIXEL_MEANS).cuda()
# tensor3/=torch.tensor(PIXEL_STDS).cuda()
# three_end=time.time()
# print('sub div in torch.tensor on cuda,time {:.4f}'.format(three_end-three_start))

# del img

#方式四 轉(zhuǎn)換成torch.tensor，使用transform方法減去均值除以方差
four_start=time.time()
img=cv2.imread(img_path)
img=img[:,:,::-1]
transform=transforms.Compose(
  [transforms.ToTensor(),transforms.Normalize(PIXEL_MEANS, PIXEL_STDS)]
)
tensor4=transform(img.copy())
four_end=time.time()
print('sub div in torch.tensor with torchvision.transforms,time {:.4f}'.format(four_end-four_start))

del img

if torch.sum(tensor1-tensor2)<=1e-3:
  print('tensor1=tensor2')
if torch.sum(tensor2-tensor4)==0:
  print('tensor2=tensor3')
# if tensor3==tensor4:
#   print('tensor3=tensor4')