def __init__(self, root, arg):
self.dataset_dir = os.path.join(root, cfg.dataset_path)
self.dataset_dir = os.path.join(self.dataset_dir, 'Train')
self.hd5_name = (
f"{cfg.hdf5_name}_d{arg.d}_path{arg.patch_size[0]}"
f"_stride{arg.stride_size[0]}_heatmap{arg.heatmap_size[0]}"
f"_version{arg.version}.hdf5")
self.hd5_path = os.path.join(self.dataset_dir, self.hd5_name)
self.transform = utils.transform(arg.version)
def test_model(image_path, data, model, arg):
device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')
img_transform = utils.transform(arg.version)
model.train(False)
_, _, _, stain, _ = steinseperation.stainsep(image_path)
H_Channel = stain[0]
cropped, coords = utils.patch_extraction(data,
H_Channel,
arg.patch_size,
arg.stride_size,
arg.version)
H, W, _ = data.shape
cell = np.zeros((H, W))
count = np.zeros((H, W))
[H_prime, W_prime] = arg.heatmap_size
for img, coord in zip(cropped, coords):
img = img_transform(np.uint8(img))
img = img.float()
img = img.to(device)
img = img.unsqueeze(0)
point, h = model(img)
heat_map = utils.heat_map_tensor(point.view(-1, 2),
h.view(-1, 1),
device,
arg.d,
arg.heatmap_size)
heatmap = heat_map.cpu().detach().numpy().reshape((H_prime, W_prime))
start_H, end_H, start_W, end_W = utils.find_out_coords(coord,
arg.patch_size,
arg.heatmap_size)
cell[start_H:end_H, start_W:end_W] += heatmap
idx = np.argwhere(heatmap != 0)
count[idx[:,0]+start_H, idx[:,1]+start_W] += 1
count[count==0] = 1
cell = np.divide(cell, count)
return cell
def __init__(self, root, arg, file):
self.imgs = []
self.coords_list = []
self.version = arg.version
self.patch_size = arg.patch_size
self.stride_size = arg.stride_size
self.transform = utils.transform(arg.version)
self.file = file
self.dataset_dir = os.path.join(root, cfg.dataset_path)
self.dataset_dir = os.path.join(self.dataset_dir, 'Test')
self.Image_path = os.path.join(self.dataset_dir, cfg.image_path)
self.Stain_path = os.path.join(self.dataset_dir, cfg.stain_path)
self.process()
def print_values(img, model, arg):
device = torch.device(
'cuda') if torch.cuda.is_available() else torch.device('cpu')
img_transform = utils.transform(arg.version)
img = img_transform(np.uint8(img))
img = img.float()
img = img.to(device)
img = img.unsqueeze(0)
model.train(False)
point, h = model(img)
print('points :', point.cpu().detach().numpy().reshape((-1, 2)))
print('H :', h.cpu().detach().numpy().reshape((-1, 1)))
print(20 * '*')
return utils.heat_map_tensor(point.view(-1, 2), h.view(-1, 1), device,
arg.d, arg.heatmap_size)
def __init__(self, root, patch_size, stride_size, d, out_size, version):
self.out_size = out_size
self.imgs = []
self.heat_maps = []
self.epsilons = 0
self.transform = utils.transform(version)
Image_path = root + cfg.dataset_path + cfg.image_path
Center_path = root + cfg.dataset_path + cfg.center_path
# MacOS thing :)
utils.delete_file(Image_path, '.DS_Store')
utils.delete_file(Center_path, '.DS_Store')
# load all image files, sorting them to
total_num = len(list(os.listdir(Image_path)))
for idx, (img_name, center_file) in enumerate(
zip(list(sorted(os.listdir(Image_path))),
list(sorted(os.listdir(Center_path))))):
# Check if the img and the centers are for same file
assert img_name[:
-4] == center_file[:
-4], "The Image {} and Center name {} are not same!".format(
img_name, center_file)
img_path = os.path.join(Image_path, img_name)
center_path = os.path.join(Center_path, center_file)
# Load Image
img = cv2.imread(img_path)
# Obtaining H-Channel
# Vahadane color normalization --> my implementation
# of Matlab's version
_, _, _, stain, _ = steinseperation.stainsep(img, 2, 0.02)
H_Channel = stain[0]
cropped, coords = utils.patch_extraction(img, H_Channel,
patch_size, stride_size,
version)
# Reading Centers
center_txt = open(center_path, "r")
center = utils.read_center_txt(center_txt)
# Extracting patches' centers
patch_centers = utils.center_extraction(center, coords, patch_size,
self.out_size)
# Finding epsilon and heatmaps
h_map, epsilon = utils.heat_map(patch_centers, coords, d,
patch_size, self.out_size)
self.imgs.extend(cropped)
self.heat_maps.extend(h_map)
self.epsilons += epsilon
print(idx + 1,
'from',
total_num,
'Images are Loaded!',
sep=' ',
end='\r',
flush=True)