Progress UNet July Oskar

What I did:

1st push: "UNet class implementation, TODO" Check if: Augmentation -> image resize shape is power of 2? Data -> original images different sizes? square? power of 2?

function: model_summary-> summary (model size, depth, etc)

class: qim_UNet -> choose between small, medium, large dataset

---

2nd push: "working on hyperparameter class" class: qim_hyperparameters -> insert model,epochs,lr,optimizer, momentum, weight decay, loss returns a 'hyper_dict' with {optimizer,criterion,n_epochs}

---

3rd push: "hyperparameter class implementation + TODO models" TODO: qim3d/utils/models.py 'preds > 0.5' -> outputs of model not in [0,1], need to use sigmoid beforehand

class: qim_hyperparameters: ordering for better understanding (_optimizer and _loss_functions)

---

4th push: "implementing training function" function: train_model -> takes model, qim_hyperparameters, trainloader,valoader. performs training loop. returns 2 dict: train_loss and val_loss each dict has: loss per epoch, and loss per batch.

minor changes to: class: qim_hyperparameters

---

5th push: "loss visualization function,creation condensed notebook" created new jupyter_notebook with condensed version

function: visualizations.py -> plt_metrics function: plots output of train_model

---

6th push: "Solution tqdm training + update both notebooks" added tqdm loop to train_model function (/!\ see issue)

---

7th push: "minor changes ValueError, log.warnings" Changed some ValueError messages from previous functions made: changed to: "invalid ___: {input}, please use: {suggestions}"

---

8th push: "removed batch" removed small test made in previous push, which wasn't removed.

docs/notebooks/Unet_condensed.ipynb

 %% Cell type:code id:be66055b-8ee9-46be-ad9d-f15edf2654a4 tags:
 ``` python
 %load_ext autoreload
 %autoreload 2
 ```
 %% Cell type:code id:0c61dd11-5a2b-44ff-b0e5-989360bbb677 tags:
 ``` python
 from os.path import join
 import numpy as np
 import torch
 import matplotlib.pyplot as plt
 from torch.utils.data import DataLoader
 import qim3d
 %matplotlib inline
 ```
 %% Cell type:code id:cd6bb832-1297-462f-8d35-1738a9c37ffd tags:
 ``` python
 # Define function for getting dataset path from string
 def get_dataset_path(name: str, datasets):
     assert name in datasets, 'Dataset name must be ' + ' or '.join(datasets)
     dataset_idx = datasets.index(name)
     datasets_path = [
         '/dtu/3d-imaging-center/projects/2023_STUDIOS_SD/analysis/data/Belialev2020/side',
         '/dtu/3d-imaging-center/projects/2023_STUDIOS_SD/analysis/data/Gaudez2022/3d',
         '/dtu/3d-imaging-center/projects/2023_STUDIOS_SD/analysis/data/Guo2023/2d/',
         '/dtu/3d-imaging-center/projects/2023_STUDIOS_SD/analysis/data/Stan2020/2d',
         '/dtu/3d-imaging-center/projects/2023_STUDIOS_SD/analysis/data/Reichardt2021/2d',
         '/dtu/3d-imaging-center/projects/2023_STUDIOS_SD/analysis/data/TestCircles/2d_binary'
     ]
     return datasets_path[dataset_idx]
 ```
 %% Cell type:code id:b8f54ff7-b3f6-4c32-87a4-b500dd877f17 tags:
 ``` python
 my_aug = qim3d.qim3d.utils.Augmentation(resize = 256)
 aug_train = my_aug.augment('heavy')
 aug_val_test = my_aug.augment(None)
 ```
 %% Cell type:code id:9a3b9c3c-4bbb-4a19-9685-f68c437e8bee tags:
 ``` python
 datasets = ['belialev2020_side', 'gaudez2022_3d', 'guo2023_2d',
             'stan2020_2d', 'reichardt2021_2d', 'testcircles_2dbinary']
-dataset = datasets[3]
+dataset = datasets[-1]
 # should not use gaudez2022: 3d image
 # reichardt2021: multiclass segmentation
 train_set = qim3d.qim3d.utils.Dataset(root_path=get_dataset_path(dataset,datasets),transform=aug_train)
 val_set = qim3d.qim3d.utils.Dataset(root_path=get_dataset_path(dataset,datasets),transform=aug_val_test)
 test_set = qim3d.qim3d.utils.Dataset(root_path=get_dataset_path(dataset,datasets),split='test',transform=aug_val_test)
 ```
 %% Output
-    images are all the same size!
-    Not implemented yet: process images that are not powers of 2.
-    images are all the same size!
-    Not implemented yet: process images that are not powers of 2.
-    images are all the same size!
-    Not implemented yet: process images that are not powers of 2.
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
+    (256, 256)
 %% Cell type:code id:4ab857f1-4595-4a21-b07b-9bdef0117c78 tags:
 ``` python
 VAL_FRACTION = 0.3
 split_idx = int(np.floor(VAL_FRACTION * len(train_set)))
 indices = torch.randperm(len(train_set))
 train_set = torch.utils.data.Subset(train_set, indices[split_idx:])
 val_set = torch.utils.data.Subset(val_set, indices[:split_idx])
 ```
 %% Cell type:code id:d94c5521-f934-4fae-b3aa-ed22546ced88 tags:
 ``` python
 # Define batch sizes
 TRAIN_BATCH_SIZE = 3
 VAL_BATCH_SIZE = 3
 TEST_BATCH_SIZE = 3
 # Define dataloaders
 train_loader = DataLoader(dataset=train_set, batch_size=TRAIN_BATCH_SIZE, shuffle=True, num_workers=8, pin_memory=True)
 val_loader = DataLoader(dataset=val_set, batch_size=VAL_BATCH_SIZE, num_workers=8, pin_memory=True)
 test_loader = DataLoader(dataset=test_set, batch_size=TEST_BATCH_SIZE, num_workers=8, pin_memory=True)
 ```
 %% Cell type:code id:ce64ae65-01fb-45a9-bdcb-a3806de8469e tags:
 ``` python
 # choosing the device to be run on
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 # defining model
-qim_unet = qim3d.qim3d.utils.qim_UNet(size = 'medium')
+qim_unet = qim3d.qim3d.models.UNet(size = 'medium')
 model = qim_unet()
 model.to(device)
 # model hyperparameters
-qim_hyper = qim3d.qim3d.utils.qim_hyperparameters(model, n_epochs=50, learning_rate = 1e-3)
+qim_hyper = qim3d.qim3d.models.Hyperparameters(model, n_epochs=50, learning_rate = 1e-3)
 hyper_dict = qim_hyper()
 # training model:
 train_loss, val_loss = qim3d.qim3d.utils.train_model(model, hyper_dict, train_loader, val_loader)
 ```
 %% Output
-    Epoch   0, train loss: 0.3357, val loss: 0.2101
-    Epoch   5, train loss: 0.0333, val loss: 0.0487
-    Epoch  10, train loss: 0.0307, val loss: 0.0244
-    Epoch  15, train loss: 0.0301, val loss: 0.0315
-    Epoch  20, train loss: 0.0171, val loss: 0.0170
-    Epoch  25, train loss: 0.0138, val loss: 0.0139
-    Epoch  30, train loss: 0.0113, val loss: 0.0096
-    Epoch  35, train loss: 0.0119, val loss: 0.0119
-    Epoch  40, train loss: 0.0077, val loss: 0.0082
-    Epoch  45, train loss: 0.0087, val loss: 0.0083
+    ---------------------------------------------------------------------------
+    KeyboardInterrupt                         Traceback (most recent call last)
+Cell     In[8], line 14
+         11 hyper_dict = qim_hyper()
+         13 # training model:
+    ---> 14 train_loss, val_loss = qim3d.qim3d.utils.train_model(model, hyper_dict, train_loader, val_loader)
+File     ~/qim3d/qim3d/utils/models.py:69, in train_model(model, qim_hyperparameters, train_loader, val_loader, eval_every, print_every)
+         66 targets = targets.to(device).type(torch.cuda.FloatTensor).unsqueeze(1)
+         68 optimizer.zero_grad()
+    ---> 69 outputs = model(inputs)
+         70 loss = criterion(outputs, targets)
+         72 # Backpropagation
+File     ~/.local/lib/python3.10/site-packages/torch/nn/modules/module.py:1501, in Module._call_impl(self, *args, **kwargs)
+       1496 # If we don't have any hooks, we want to skip the rest of the logic in
+       1497 # this function, and just call forward.
+       1498 if not (self._backward_hooks or self._backward_pre_hooks or self._forward_hooks or self._forward_pre_hooks
+       1499         or _global_backward_pre_hooks or _global_backward_hooks
+       1500         or _global_forward_hooks or _global_forward_pre_hooks):
+    -> 1501     return forward_call(*args, **kwargs)
+       1502 # Do not call functions when jit is used
+       1503 full_backward_hooks, non_full_backward_hooks = [], []
+File     ~/miniconda3/lib/python3.10/site-packages/monai/networks/nets/unet.py:303, in UNet.forward(self, x)
+        302 def forward(self, x: torch.Tensor) -> torch.Tensor:
+    --> 303     x = self.model(x)
+        304     return x
+File     ~/.local/lib/python3.10/site-packages/torch/nn/modules/module.py:1501, in Module._call_impl(self, *args, **kwargs)
+       1496 # If we don't have any hooks, we want to skip the rest of the logic in
+       1497 # this function, and just call forward.
+       1498 if not (self._backward_hooks or self._backward_pre_hooks or self._forward_hooks or self._forward_pre_hooks
+       1499         or _global_backward_pre_hooks or _global_backward_hooks
+       1500         or _global_forward_hooks or _global_forward_pre_hooks):
+    -> 1501     return forward_call(*args, **kwargs)
+       1502 # Do not call functions when jit is used
+       1503 full_backward_hooks, non_full_backward_hooks = [], []
+File     ~/.local/lib/python3.10/site-packages/torch/nn/modules/container.py:217, in Sequential.forward(self, input)
+        215 def forward(self, input):
+        216     for module in self:
+    --> 217         input = module(input)
+        218     return input
+File     ~/.local/lib/python3.10/site-packages/torch/nn/modules/module.py:1501, in Module._call_impl(self, *args, **kwargs)
+       1496 # If we don't have any hooks, we want to skip the rest of the logic in
+       1497 # this function, and just call forward.
+       1498 if not (self._backward_hooks or self._backward_pre_hooks or self._forward_hooks or self._forward_pre_hooks
+       1499         or _global_backward_pre_hooks or _global_backward_hooks
+       1500         or _global_forward_hooks or _global_forward_pre_hooks):
+    -> 1501     return forward_call(*args, **kwargs)
+       1502 # Do not call functions when jit is used
+       1503 full_backward_hooks, non_full_backward_hooks = [], []
+File     ~/.local/lib/python3.10/site-packages/torch/nn/modules/container.py:217, in Sequential.forward(self, input)
+        215 def forward(self, input):
+        216     for module in self:
+    --> 217         input = module(input)
+        218     return input
+File     ~/.local/lib/python3.10/site-packages/torch/nn/modules/module.py:1501, in Module._call_impl(self, *args, **kwargs)
+       1496 # If we don't have any hooks, we want to skip the rest of the logic in
+       1497 # this function, and just call forward.
+       1498 if not (self._backward_hooks or self._backward_pre_hooks or self._forward_hooks or self._forward_pre_hooks
+       1499         or _global_backward_pre_hooks or _global_backward_hooks
+       1500         or _global_forward_hooks or _global_forward_pre_hooks):
+    -> 1501     return forward_call(*args, **kwargs)
+       1502 # Do not call functions when jit is used
+       1503 full_backward_hooks, non_full_backward_hooks = [], []
+File     ~/.local/lib/python3.10/site-packages/torch/nn/modules/conv.py:463, in Conv2d.forward(self, input)
+        462 def forward(self, input: Tensor) -> Tensor:
+    --> 463     return self._conv_forward(input, self.weight, self.bias)
+File     ~/.local/lib/python3.10/site-packages/torch/nn/modules/conv.py:459, in Conv2d._conv_forward(self, input, weight, bias)
+        455 if self.padding_mode != 'zeros':
+        456     return F.conv2d(F.pad(input, self._reversed_padding_repeated_twice, mode=self.padding_mode),
+        457                     weight, bias, self.stride,
+        458                     _pair(0), self.dilation, self.groups)
+    --> 459 return F.conv2d(input, weight, bias, self.stride,
+        460                 self.padding, self.dilation, self.groups)
+    KeyboardInterrupt:
 %% Cell type:code id:27fb326d-2771-470d-8288-a91d71e38fce tags:
 ``` python
 plt.figure(figsize=(16, 6))
 qim3d.qim3d.viz.plot_metrics(train_loss, label = 'Train')
 qim3d.qim3d.viz.plot_metrics(val_loss,color = 'orange', label = 'Valid.')
 plt.show()
 ```
 %% Output
 %% Cell type:code id:f8684cb0-5673-4409-8d22-f00b7d099ca4 tags:
 ``` python
 in_targ_preds_test = qim3d.utils.inference(test_set,model)
 qim3d.qim3d.viz.grid_pred(in_targ_preds_test,alpha=1)
 ```
 %% Output