draft3.py

import random
import os
from PIL import Image,ImageFilter,ImageDraw
import numpy as np
import h5py
from PIL import ImageStat
import cv2
import torchvision, torch 
from torchvision import datasets, transforms
import glob
from PIL import Image
import h5py
from matplotlib import pyplot as plt
import json

# with open('/home/nvhuy/hangdtth/hang/CSR_KD/trancos_training.json', 'r') as outfile:        
#     train_list = json.load(outfile)

# # with open('/home/nvhuy/hangdtth/hang/CSR_KD/shanghai.json', 'r') as outfile:        
# #     train_list = json.load(outfile)

    
# train_loader = dataset.listDataset(train_list,
#                     shuffle=True,
#                     transform=transforms.Compose([
#                                         transforms.ToTensor(),
#                                         transforms.Normalize(mean=[0.485, 0.456, 0.406], # from img net 
#                                                         std=[0.229, 0.224, 0.225]),
#                     ]), 
#                     train=True, 
#                     # seen=model.seen,
#                     batch_size=2,
#                     num_workers=4)                     


# for i, (img, target) in enumerate(train_loader):
    
#     # print(f"img: {img}\ntarget: {target}")
#     print(f"target shape: {target.shape}")
#     print(f"target sum: {target.sum()}")
#     if i == 3 : 
#         break

# import glob 
# l = []
# for f in glob.glob("/home/nvhuy/hangdtth/hang/ShanghaiTech/part_A/train_data/images/*"): 
#     l.append(f)

# # print(l)
# with open('shanghai.json', 'w') as outfile:
#     json.dump(l, outfile)


# img_path = "/home/nvhuy/hangdtth/hang/ShanghaiTech/part_A/train_data/images/IMG_1.jpg"
# gt_path = img_path.replace('.jpg','.h5').replace('images','ground_truth')
# img = Image.open(img_path).convert('RGB')
# gt_file = h5py.File(gt_path, 'r')
# target = np.asarray(gt_file['density'])
# print(f"Target size 1 : {target.shape}")
# target = cv2.resize(target,(int(target.shape[1]/8), int(target.shape[0]/8)),interpolation = cv2.INTER_CUBIC)*64
# print(f"Target size 2 : {target.shape}")

# target = cv2.resize(target,(int(target.shape[1]/8), int(target.shape[0]/8)),interpolation = cv2.INTER_CUBIC)
# print(f"Target size 3 : {target.shape}")



# _path = "/home/nvhuy/hangdtth/hang/ShanghaiTech/part_A/test_data"
# ls = []
# for file in glob.glob(_path+"/images/*"): 
#     # img = Image.open(file)
#     # gt_path = file.replace('images', 'ground_truth_h5').replace('jpg', 'h5')
#     # gt_ = h5py.File(gt_path, 'r')
#     # gt = gt_['density']
#     # print(type(gt.value))
#     # # plt.imshow(img, alpha=.5)
#     # # plt.imshow(gt, alpha=.85)
#     # # plt.show()
#     # break
#     ls.append(file)

# with open('shanghai_test.json','w') as outfile:
#     json.dump(ls, outfile)
'''
a
 tensor([[[[ 1.,  2.,  3.,  4.],
          [ 5.,  6.,  7.,  8.]],

         [[ 9., 10., 11., 12.],
          [13., 14., 15., 16.]]],


        [[[17., 18., 19., 20.],
          [21., 22., 23., 24.]],

         [[25., 26., 27., 28.],
          [29., 30., 31., 32.]]]]) 

b
 tensor([[[[33., 34., 35., 36.],
          [37., 38., 39., 40.]],

         [[41., 42., 43., 44.],
          [45., 46., 47., 48.]]],


        [[[49., 50., 51., 52.],
          [53., 54., 55., 56.]],

         [[57., 58., 59., 60.],
          [61., 62., 63., 64.]]]]) 




f1: 
tensor([[[ 1,  2,  3,  4,  5,  6,  7,  8],
         [ 9, 10, 11, 12, 13, 14, 15, 16]],

        [[17, 18, 19, 20, 21, 22, 23, 24],
         [25, 26, 27, 28, 29, 30, 31, 32]]])

f2: tensor([[[33, 34, 35, 36, 37, 38, 39, 40],
         [41, 42, 43, 44, 45, 46, 47, 48]],

        [[49, 50, 51, 52, 53, 54, 55, 56],
         [57, 58, 59, 60, 61, 62, 63, 64]]])

f2: tensor([[[33, 41],
         [34, 42],
         [35, 43],
         [36, 44],
         [37, 45],
         [38, 46],
         [39, 47],
         [40, 48]],

        [[49, 57],
         [50, 58],
         [51, 59],
         [52, 60],
         [53, 61],
         [54, 62],
         [55, 63],
         [56, 64]]])

FSP_matrix:
 tensor([[[ 1356,  1644],
         [ 3692,  4492]],

        [[ 8652,  9964],
         [12012, 13836]]])          
    

a = "bc"
b = "cd"
e = "bc"

if a == b :  
  print("a = e")
else :
  print("a =! e")
'''      


s_out = torch.randn(15).view(1,1,3,5)
target = torch.randn(20).view(1,1,4,5)
print("\n1)__________\n")
print(s_out,"\n", target)

# _h_min = torch.min(s_out.shape[2], target.shape[2]) + 1
# _w_min = torch.min(s_out.shape[3], target.shape[3]) + 1

_h_min = s_out.shape[2] if s_out.shape[2] < target.shape[2] else target.shape[2]
_w_min = s_out.shape[3] if s_out.shape[3] < target.shape[3] else target.shape[3]


print("2)____________\n")
print("_h_min: ", _h_min)
print("_w_min: ", _w_min)
print(s_out[0,0,:_h_min,:_w_min], "\n", target[0,0,:_h_min,:_w_min])