cell-data-loader 0.0.8__tar.gz → 0.0.14__tar.gz

{cell_data_loader-0.0.8 → cell_data_loader-0.0.14}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: cell_data_loader
-Version: 0.0.8
+Version: 0.0.14
 Summary: Converts general images of cells into formats and labels for deep learning pipelines
 Project-URL: Homepage, https://github.com/mleming/CellDataLoader
 Author-email: Matt Leming <mleming@mgh.harvard.edu>
@@ -14,8 +14,8 @@ Requires-Dist: pillow
 Requires-Dist: scikit-image
 Requires-Dist: scipy
 Requires-Dist: slideio==2.4.1
-Requires-Dist: torch
-Requires-Dist: torchvision<=0.15.0
+Requires-Dist: torch<=2.0.0
+Requires-Dist: torchvision==0.15.0
 Description-Content-Type: text/markdown
 
 Cell Data Loader

{cell_data_loader-0.0.8 → cell_data_loader-0.0.14}/example/example_numpy.py

@@ -1,5 +1,5 @@
 #!/usr/bin/python3
-from src.cell_data_loader import CellDataloader
+from cell_data_loader import CellDataloader
 
 try:
 	import tensorflow as tf

{cell_data_loader-0.0.8 → cell_data_loader-0.0.14}/example/example_torch.py

@@ -1,6 +1,6 @@
 #!/usr/bin/python3
 
-from src.cell_data_loader import CellDataloader
+from cell_data_loader import CellDataloader
 from torchvision.models import resnet50 #, ResNet50_Weights
 import torch
 import os

{cell_data_loader-0.0.8 → cell_data_loader-0.0.14}/pyproject.toml

@@ -2,13 +2,13 @@
 
 name = "cell_data_loader"
 
-version = "0.0.8"
+version = "0.0.14"
 
 dependencies = [
   "numpy",
   "scipy",
-  "torch",
-  "torchvision<=0.15.0",
+  "torch<=2.0.0",
+  "torchvision==0.15.0",
   "opencv-python>=4.5.4",
   "slideio==2.4.1",
   "scikit-image",

{cell_data_loader-0.0.8 → cell_data_loader-0.0.14}/src/cell_data_loader/__init__.py

@@ -1,3 +1,3 @@
-__version__ = "0.1.0"
+__version__ = "0.0.14"
 
 from .cell_data_loader import *

{cell_data_loader-0.0.8 → cell_data_loader-0.0.14}/tests/unit_tests.py

@@ -139,7 +139,7 @@ class TestSimple(unittest.TestCase):
 			print("Torch import failed -- skipping test")
 			valid_torch = False
 		if valid_torch:
-			from example_torch import example_torch
+			from example.example_torch import example_torch
 			example_torch()
 	def test_label_match(self):
 		batch_size = 1

cell_data_loader-0.0.8/example_numpy.py

@@ -1,71 +0,0 @@
-#!/usr/bin/python3
-from src.cell_data_loader import CellDataloader
-
-try:
-	import tensorflow as tf
-except:
-	try:
-		from keras import tensorflow as tf
-	except:
-		raise Exception("Need valid tensorflow")
-
-import numpy as np
-import os
-
-def example_numpy(verbose=True):
-
-	"""
-	Replace these folders with whatever folders of cells you may have. Note that
-	the test/train folders in this script are the same -- you would need to 
-	have a separate train/test set in your own implementation
-	"""
-	wd = os.path.dirname(os.path.dirname(os.path.realpath(__file__)))
-	imfolder1 = os.path.join(wd,'data',
-		'3368914_4_non_tumor')
-	imfolder2 = os.path.join(wd,'data',
-		'4173633_5')
-
-	#model = resnet50()
-	model = tf.keras.applications.resnet50.ResNet50(
-		include_top=True,
-		weights='imagenet',
-		input_tensor=None,
-		input_shape=None,
-		pooling=None,
-		classes=1000
-	)
-
-	# Train
-
-	dataloader_train = CellDataloader(imfolder1,imfolder2,
-		dtype="numpy",verbose=False)
-
-	z = None
-	for epoch in range(1):
-		for image,y in dataloader_train:
-			if z is None:
-				z_dim = list(y.shape)
-				z_dim[1] = 1000 - z_dim[1]
-				z = np.zeros(z_dim)
-			y = np.concatenate((y,z),axis=1)
-			model.fit(image,y)
-
-	# Test
-
-	model.eval()
-	dataloader_test = CellDataloader(imfolder1,imfolder2,dtype="torch",
-		verbose=False)
-	total_images = 0
-	sum_accuracy = 0
-	for image,y in dataloader_test:
-		total_images += image.shape[0]
-		y_pred = model(image)
-		y_pred = y_pred[:,:y.shape[1]]
-		sum_accuracy += torch.sum(torch.argmax(y_pred,axis=1) == \
-			torch.argmax(y,axis=1))
-
-	accuracy = sum_accuracy / total_images
-	if verbose: print("Final accuracy: %.4f" % sum_accuracy)
-
-if __name__ == "__main__":
-	example_numpy()

cell_data_loader-0.0.8/example_torch.py

@@ -1,88 +0,0 @@
-#!/usr/bin/python3
-
-from src.cell_data_loader import CellDataloader
-from torchvision.models import resnet50 #, ResNet50_Weights
-import torch
-import os
-import glob
-
-def example_torch(gpu_ids = None,verbose=True):
-
-	"""
-	Replace these folders with whatever folders of cells you may have. Note that
-	the test/train folders in this script are the same -- you would need to 
-	have a separate train/test set in your own implementation
-	"""
-
-	wd = os.path.dirname(os.path.dirname(os.path.realpath(__file__)))
-	model_folder = os.path.join(wd,'checkpoints')
-	os.makedirs(model_folder,exist_ok=True)
-	model_file = os.path.join(model_folder,'torch_model.pt')
-	
-	# Training data folders
-	imfolder1_train = os.path.join(wd,'data',
-		'3368914_4_non_tumor')
-	imfolder2_train = os.path.join(wd,'data',
-		'4173633_5')
-	
-	# Test data folders
-	imfolder1_test = os.path.join(wd,'data','3368914_4_non_tumor')
-	imfolder2_test = os.path.join(wd,'data','4173633_5')
-	
-	# Get a pretrained model from torchvision
-	if os.path.isfile(model_file):
-		if verbose: print("Loading %s" % model_file)
-		model = torch.load(model_file)
-	else:
-		model = resnet50(pretrained=True)
-
-	if gpu_ids is not None:
-		model.to(gpu_ids)
-	
-	# Train
-	
-	model.train()
-	loss_fn = torch.nn.MSELoss(reduction='sum')
-	optimizer = torch.optim.Adam(model.parameters(), lr=1e-5)
-	dataloader_train = CellDataloader(imfolder1_train,imfolder2_train,
-		dtype="torch",
-		verbose=False, gpu_ids=gpu_ids)
-	if verbose: print("Beginning training")
-	n_epochs = 1
-	for epoch in range(n_epochs):
-		l = 0
-		c = 0
-		for image,y in dataloader_train:
-			y_pred = model(image)
-			y_pred = y_pred[:,:y.size()[1]]
-			loss = loss_fn(y_pred, y)
-			l += float(loss)
-			c += 1
-			optimizer.zero_grad()
-			loss.backward()
-			optimizer.step()
-		torch.save(model,model_file)
-		if verbose:
-			print(
-				"Epoch {epoch:d}/{n_epochs:d}: loss: {loss:.5f}".format(
-					epoch=epoch,n_epochs=n_epochs,loss=l/c)
-			)
-	# Test
-	
-	model.eval()
-
-	dataloader_test = CellDataloader(imfolder1_test,imfolder2_test,
-		dtype="torch",verbose=False, gpu_ids = gpu_ids)
-	total_images = 0
-	sum_accuracy = 0
-	for image,y in dataloader_test:
-		total_images += image.size()[0]
-		y_pred = model(image)
-		y_pred = y_pred[:,:y.size()[1]]
-		sum_accuracy += torch.sum(torch.argmax(y_pred,axis=1) == \
-			torch.argmax(y,axis=1))
-	accuracy = sum_accuracy / total_images
-	if verbose: print("Final accuracy: %.4f" % accuracy)
-
-if __name__ == "__main__":
-	example_torch()