def __getitem__(self, index):
sample = self.imgs[index]
if self.phase == 'train':
splits = sample.split('/', 7)[-1] #分割6次，可变
label = int(splits.split('_')[0])
img_path = sample
data = Image.open(img_path).convert('RGB')
# data = amaugimg(data)
data = self.transforms(data)
return data, label
else: #对于测试
data = Image.open(sample)
data = data.convert('RGB')
img1 = data #原图
img2 = data.transpose(Image.FLIP_LEFT_RIGHT) #水平反转图
data = data.convert('L') #灰度图
img3 = cv2.cvtColor(np.asarray(data), cv2.COLOR_RGB2BGR)
img3 = Image.fromarray(cv2.cvtColor(img3, cv2.COLOR_BGR2RGB))
img4 = img3.transpose(Image.FLIP_LEFT_RIGHT) #灰度反转
img1 = self.transforms(img1)
img2 = self.transforms(img2)
img3 = self.transforms(img3)
img4 = self.transforms(img4)
return img1, img2
def __getitem__(self, index):
#print(self.dict_labels)
#print(lst)
img_path = self.lst[index]
#dog->1,cat->0
label = self.lst[index][1]
data = imageio.imread(self.root + '/images/' + img_path[0] + '.jpg')
try:
data = self.augmentations.augment_image(data)
imageio.imwrite(
r'F:/5.datasets/test/aug_images/' + img_path[0] +
'-aug_pipline.jpg', data)
data = Image.open(r'F:/5.datasets/test/aug_images/' + img_path[0] +
'-aug_pipline.jpg')
except Exception as e:
print(e)
finally:
data = Image.open(self.root + '/images/' + img_path[0] + '.jpg')
#data = data.convert('RGB')
#array = np.asarray(pil_img)
#data = t.from_numpy(array)
#plt.imshow(data)
#plt.imshow(data)
#data = Image.open(r'F:/5.datasets/test/aug_images/'+img_path[0] + '-aug_pipline.jpg')
data = data.convert('RGB')
if self.transforms:
#data = self.transforms(TF.to_tensor(data))
data = self.transforms(data)
#data = TF.to_tensor(self.augmentations.augment_image(data))
return data, label
def __getitem__(self, index):
data = Image.open(os.path.join(self.file_path, self.ims[index]))
bic_im = self.trans_bic(data)
data = data.convert("YCbCr")
data_y, cb, cr = data.split()
data = self.trans(data_y)
# batch must contain tensors, numbers, dicts or lists;
# data_dict = {'bic':bic_im,'name':self.ims[index]}
return data, bic_im, self.ims[index]
def __getitem__(self, index):
sample = self.imgs[index]
splits = sample.split()
img_path = splits[0]
data = Image.open(img_path)
data = data.convert('L')
data = self.transforms(data)
label = np.int32(splits[1])
return data.float(), label
def __getitem__(self, index):
sample = self.imgs[index]
splits = sample.split()
img_path = os.path.join(self.root, splits[0])
data = Image.open(img_path)
data = data.convert('L')
data = self.transforms(data)
image1 = data
img_path = os.path.join(self.root, splits[1])
data = Image.open(img_path)
data = data.convert('L')
data = self.transforms(data)
image2 = data
img_path = os.path.join(self.pd_root, splits[0])
pdimg = Image.open(img_path)
pdimg = pdimg.convert('L')
pdimg1 = self.pd_transforms(pdimg)
img_path = os.path.join(self.pd_root, splits[1])
pdimg = Image.open(img_path)
pdimg = pdimg.convert('L')
pdimg2 = self.pd_transforms(pdimg)
pdimg1 = pdimg1[:, 1::4, 1::4]
pdimg2 = pdimg2[:, 1::4, 1::4]
pdimg1[pdimg1 > 165 / 256] = pdimg1[pdimg1 > 165 / 256] - 1
pdimg1[pdimg1 > 90 / 256] = 180 / 256
pdimg1 = torch.cat((torch.cos(pdimg1 * 256 / 180 * math.pi), torch.sin(pdimg1 * 256 / 180 * math.pi)), 0)
pdimg2[pdimg2 > 165 / 256] = pdimg2[pdimg2 > 165 / 256] - 1
pdimg2[pdimg2 > 90 / 256] = 180 / 256
pdimg2 = torch.cat((torch.cos(pdimg2 * 256 / 180 * math.pi), torch.sin(pdimg2 * 256 / 180 * math.pi)), 0)
dx = np.int32(splits[2])
dy = np.int32(splits[3])
da = np.int32(splits[4])
label = torch.FloatTensor([dx/25.0, dy/25.0, da/20])
return image1.float(), image2.float(), (pdimg1.float(), pdimg2.float()), label
def __getitem__(self, index):
data = self.index_to_data[index]
img_path, label = data[self.img_colname], data[self.label_colname]
if self.relative_path:
img_path = os.path.join(self.root_path, img_path)
data = Image.open(img_path)
data = data.convert(self.img_to)
data = self.transforms(data)
return data, label
def __getitem__(self, index):
img_path = self.imgs[index]
label = int(img_path.strip().split('/')[10])
print(img_path, label)
#data = Image.open(img_path)
data = io.imread(img_path)
data = Image.fromarray(data)
if data.getbands()[0] == 'L':
data = data.convert('RGB')
data = self.transforms(data)
return data, label
def __getitem__(self, index):
"""
一次返回一张图片的数据
"""
img_path = self.root + str(self.imgs[index]) + '.jpg'
data = Image.open(img_path)
data = data.convert('RGB')
data = self.transforms(data)
label = self.labels[index]
return data, label
def __getitem__(self, index):
ImageFile.LOAD_TRUNCATED_IMAGES = True
img_path = self.imgs[index]
if self.test:
label = img_path.split('\\')[-1]
else:
label = int(img_path.split('\\')[-2]) - 1
with Image.open(img_path) as data:
data = data.convert('RGB')
data = self.transforms(data)
return data, label
def __getitem__(self, index):
"""
一次返回一张图片的数据
"""
img_path = self.imgs[index]
label = self.labels[index]
# img = cv2.imread(img_path)
# img = cv2.cvtColor(img,cv2.COLOR_BGR2RGB)
data = Image.open(img_path)
data = data.convert("RGB") # 如果有4通道图片转化为3通道
data = self.transforms(data)
return data, label, img_path, opt.cate_classes[label] # 返回数据级标签图片路径
def __getitem__(self, index):
sample = self.imgs[index]
splits = sample.split()
img_path = os.path.join(self.root, splits[0])
data = Image.open(img_path)
data = data.convert('L')
data = self.transforms(data)
image1 = data
img_path = os.path.join(self.root, splits[1])
data = Image.open(img_path)
data = data.convert('L')
data = self.transforms(data)
image2 = data
dx = np.int32(splits[2])
dy = np.int32(splits[3])
da = np.int32(splits[4])
label = torch.FloatTensor([dx/50.0, dy/50.0, da/40])
return image1.float(), image2.float(), label
def __getitem__(self, index):
sample = self.imgs[index]
if self.phase == 'train':
splits = sample.split('/')[-1]
label = int(splits.split('_')[0])
img_path = sample
data = Image.open(img_path).convert('RGB')
data = self.transforms(data)
return data, label
else:
data = Image.open(sample)
data = data.convert('RGB')
data = self.transforms(data)
return data
def __getitem__(self, index):
#print(self.dict_labels)
#print(lst)
img_path = self.lst[index]
#dog->1,cat->0
label = self.lst[index][1]
data = Image.open(self.root + '/images/' + img_path[0] + '.jpg')
data = data.convert('RGB')
#array = np.asarray(pil_img)
#data = t.from_numpy(array)
if self.transforms:
data = self.transforms(data)
return data, label
def __getitem__(self, index):
'''
return the data of one image
'''
data_path = self.images_root + 'testimg/' + self.images_path[index]
data = Image.open(data_path)
data = data.convert('RGB')
Data = self.transforms_data(data)
# *Img.getcolors()
# Label = self.transforms_label(label)
return Data, str(self.images_path[index])
def __getitem__(self, index):
sample = self.imgs[index]
if self.phase == 'train':
splits = sample.split('/', 6)[-1] #分割6次，可变
label = self.dic[splits]
img_path = sample
data = Image.open(img_path)
# data = data.convert('L')
data = self.transforms(data)
return data.float(), label
else:
data = Image.open(sample)
data = data.convert('RGB') #<-----------------------
data = self.transforms(data)
return data
def read_throw(self, index):
splits = self.imgs[index]
bb = self.bbs[index]
img_path = splits[0]
data = Image.open(img_path)
data.load()
data = data.convert('RGB')
data = data.crop(self.refine_bb(bb, data.size))
if self.transforms is not None:
data = self.transforms(data)
label = int(splits[-1])
return data, label
def __getitem__(self, index):
sample = self.imgs[index]
splits = sample.split()
img_path = splits[0]
data = Image.open(img_path)
data = data.convert('RGB')
data = self.rand_augment(data)
data = self.transforms(data)
label = np.int32(splits[1])
if label > (self.NUM_CLASSES - 1):
label = (self.NUM_CLASSES - 1)
levels = [1] * label + [0] * (self.NUM_CLASSES - 1 - label)
levels = torch.tensor(levels, dtype=torch.float32)
return data, label, levels
def __getitem__(self, index):
sample_path = self.imgs[index] # 当前图片的路径
sample_dir = os.path.dirname(sample_path) # 当前图片的上级目录的路径
# 标签必须是从0~classNums的连续整数
class_name = sample_dir.split(self.path_spilt)[-1] # 获取类名(当前图片的文件夹名字)
label = self.classes.index(class_name) # 根据类名从列表里找到它的索引, 把索引号当作标签
data = Image.open(sample_path) # Image打开返回的是RGB (H , W , C)
if self.input_shape[0] is 1: # 输入是1维, 则转成灰度图
data = data.convert(
'L') # 转为灰度图像(1维), 公式L = R*0.299 + G*0.587+ B*0.114
data = self.transforms(data)
return data.float(), label
def __getitem__(self, index):
"""
一次返回一张图片的数据
"""
img_path = self.imgs[index]
if self.test:
# label = self.imgs[index].split('.')[-2].split('/')[-1]
label = img_path.split('/')[-1]
else:
label = self.labels[index]
data = Image.open(img_path)
if opt.gray == True:
dataRGB = data.convert('RGB')
dataRGB = self.transforms(dataRGB)
return dataRGB, label
data = self.transforms(data)
return data, label
def __getitem__(self, index):
sample = self.imgs[index]
splits = sample.split()
img_path = splits[0]
data = Image.open(img_path)
if self.input_shape[0] == 1: # channel = 1
data = data.convert('L')
data = self.transforms(data)
# data_view = data.cpu().numpy().transpose(1, 2, 0)
# plt.imshow(data_view)
# # plt.axis('off')
# plt.savefig(f'datasets/test/test_m{index}.jpg', bbox_inches='tight')
# plt.show()
if self.phase == 'test':
return data.float()
else:
label = np.int32(splits[1])
return data.float(), label
def __getitem__(self, index):
img_path = self.imgs[index]
label = self.labels[index]
# 提取特征
if self.extract_feature:
label = self.imgs[index].rsplit('/', 1)[-1]
try:
data = Image.open(img_path)
except Exception as e:
print(e)
else:
if len(data.split())!=3:
data = data.convert('RGB')
#print(len(data.split()))
if self.transforms:
data = self.transforms(data)
return data, label
return torch.zeros(3, 224, 224), 0
