From 658b5009f9d69ac6a881d31abb5301f81fffa338 Mon Sep 17 00:00:00 2001
From: "s.islam" <s.islam@fz-juelich.de>
Date: Mon, 25 Apr 2022 01:33:16 +0200
Subject: [PATCH] RMI Loss added and 3 channel code(commented/not functional)
added.
---
code/data.py | 95 ++++++++++++++++++++++++++++++++++++++++++++++++---
code/model.py | 5 +--
2 files changed, 93 insertions(+), 7 deletions(-)
diff --git a/code/data.py b/code/data.py
index 235cae1..40c4d76 100644
--- a/code/data.py
+++ b/code/data.py
@@ -121,14 +121,15 @@ class TestDataModule(pl.LightningDataModule):
# TODO: Load the PLI and Cytp train data here as lists of numpy arrays: List[np.ndarray]
# Load the pyramid/00 per file
+ #For single channel
#For JSC Training.
- pli_path = '/p/fastdata/pli/Private/oberstrass1/datasets/vervet1818/vervet1818-stained/data/aligned/pli/NTransmittance'
- cyto_path = '/p/fastdata/pli/Private/oberstrass1/datasets/vervet1818/vervet1818-stained/data/aligned/stained'
+ #pli_path = '/p/fastdata/pli/Private/oberstrass1/datasets/vervet1818/vervet1818-stained/data/aligned/pli/NTransmittance'
+ #cyto_path = '/p/fastdata/pli/Private/oberstrass1/datasets/vervet1818/vervet1818-stained/data/aligned/stained'
#For Local Machine Training.
- #pli_path = '/media/tushar/A2246889246861F1/Master Thesis MAIA/example-data/pli/NTransmittance'
- #cyto_path = '/media/tushar/A2246889246861F1/Master Thesis MAIA/example-data/stained'
-
+ pli_path = '/media/tushar/A2246889246861F1/Master Thesis MAIA/example-data/pli/NTransmittance'
+ cyto_path = '/media/tushar/A2246889246861F1/Master Thesis MAIA/example-data/stained'
+
pli_files_list = [file for file in os.listdir(pli_path) if
file.endswith(('.h5', '.hdf', '.h4', '.hdf4', '.he2', '.hdf5', '.he5'))]
pli_files_list.sort()
@@ -145,6 +146,58 @@ class TestDataModule(pl.LightningDataModule):
pli_train_file = np.asarray(pli_train_file).astype(np.float32)
pli_train_file = pli_train_file - 0.5
self.pli_train.append(pli_train_file)
+
+
+ # For 3 Channel.
+ '''
+ pli_NTransmittance_path = '/media/tushar/A2246889246861F1/Master Thesis MAIA/example-data/pli/NTransmittance'
+ pli_Retardation_path = '/media/tushar/A2246889246861F1/Master Thesis MAIA/example-data/pli/Retardation'
+ pli_Direction_path = '/media/tushar/A2246889246861F1/Master Thesis MAIA/example-data/pli/Direction'
+ cyto_path = '/media/tushar/A2246889246861F1/Master Thesis MAIA/example-data/stained'
+
+ pli_NTransmittance_files_list = [file for file in os.listdir(pli_NTransmittance_path) if
+ file.endswith(('.h5', '.hdf', '.h4', '.hdf4', '.he2', '.hdf5', '.he5'))]
+ pli_NTransmittance_files_list.sort()
+
+ pli_Retardation_files_list = [file for file in os.listdir(pli_Retardation_path) if
+ file.endswith(('.h5', '.hdf', '.h4', '.hdf4', '.he2', '.hdf5', '.he5'))]
+ pli_Retardation_files_list.sort()
+
+ pli_Direction_files_list = [file for file in os.listdir(pli_Direction_path) if
+ file.endswith(('.h5', '.hdf', '.h4', '.hdf4', '.he2', '.hdf5', '.he5'))]
+ pli_Direction_files_list.sort()
+ cyto_files_list = [file for file in os.listdir(cyto_path) if
+ file.endswith(('.h5', '.hdf', '.h4', '.hdf4', '.he2', '.hdf5', '.he5'))]
+
+ pli_NTransmittance_train = []
+ pli_Retardation_train = []
+ pli_Direction_train = []
+
+ for i in range(0, 4):
+ pli_train_file = h5py.File(os.path.join(pli_NTransmittance_path, pli_NTransmittance_files_list[i]), 'r')
+ pli_train_file = pli_train_file['pyramid/00']
+ pli_train_file = np.asarray(pli_train_file).astype(np.float32)
+ pli_train_file = pli_train_file - 0.5
+ pli_NTransmittance_train.append(pli_train_file)
+ for i in range(0, 4):
+ pli_train_file = h5py.File(os.path.join(pli_Retardation_path, pli_Retardation_files_list[i]), 'r')
+ pli_train_file = pli_train_file['pyramid/00']
+ pli_train_file = np.asarray(pli_train_file).astype(np.float32)
+ pli_train_file = pli_train_file - 0.5
+ pli_Retardation_train.append(pli_train_file)
+ for i in range(0, 4):
+ pli_train_file = h5py.File(os.path.join(pli_Direction_path, pli_Direction_files_list[i]), 'r')
+ pli_train_file = pli_train_file['pyramid/00']
+ pli_train_file = np.asarray(pli_train_file).astype(np.float32)
+ pli_train_file = (pli_train_file/255) - 0.5
+ pli_Direction_train.append(pli_train_file)
+
+ self.pli_train = []
+ self.cyto_train = []
+
+ for i in range(0,4):
+ self.pli_train.append(np.stack((pli_NTransmittance_train[i], pli_Retardation_train[i], pli_Direction_train[i]), axis=-1))
+ '''
for i in range(0,4):
cyto_train_file = h5py.File(os.path.join(cyto_path, cyto_files_list[i]), 'r')
@@ -165,12 +218,44 @@ class TestDataModule(pl.LightningDataModule):
self.pli_val = []
self.cyto_val = []
+
+ #Single Channel
pli_val_file = h5py.File(os.path.join(pli_path, pli_files_list[4]), 'r')
pli_val_file = pli_val_file['pyramid/00']
pli_val_file = np.asarray(pli_val_file).astype(np.float32)
pli_val_file = pli_val_file - 0.5
self.pli_val.append(pli_val_file)
+ #3 Channels
+ '''
+ pli_NTransmittance_val = []
+ pli_Retardation_val = []
+ pli_Direction_val = []
+
+
+ pli_NTransmittance_val_file = h5py.File(os.path.join(pli_NTransmittance_path, pli_NTransmittance_files_list[4]), 'r')
+ pli_NTransmittance_val_file = pli_NTransmittance_val_file['pyramid/00']
+ pli_NTransmittance_val_file = np.asarray(pli_NTransmittance_val_file).astype(np.float32)
+ pli_NTransmittance_val_file = pli_NTransmittance_val_file - 0.5
+ pli_NTransmittance_val.append(pli_NTransmittance_val_file)
+
+ pli_Retardation_val_file = h5py.File(os.path.join(pli_Retardation_path, pli_Retardation_files_list[4]), 'r')
+ pli_Retardation_val_file = pli_Retardation_val_file['pyramid/00']
+ pli_Retardation_val_file = np.asarray(pli_Retardation_val_file).astype(np.float32)
+ pli_Retardation_val_file = pli_Retardation_val_file - 0.5
+ pli_Retardation_val.append(pli_Retardation_val_file)
+
+ pli_Direction_val_file = h5py.File(os.path.join(pli_Direction_path, pli_Direction_files_list[4]), 'r')
+ pli_Direction_val_file = pli_Direction_val_file['pyramid/00']
+ pli_Direction_val_file = np.asarray(pli_Direction_val_file).astype(np.float32)
+ pli_Direction_val_file = (pli_Direction_val_file/255) - 0.5
+ pli_Direction_val.append(pli_Direction_val_file)
+
+ self.pli_val = np.stack((pli_NTransmittance_val, pli_Retardation_val, pli_Direction_val), axis=-1)
+ '''
+
+
+
cyto_val_file = h5py.File(os.path.join(cyto_path, cyto_files_list[4]), 'r')
cyto_val_file = cyto_val_file['pyramid/00']
cyto_val_file = np.asarray(cyto_val_file).astype(np.float32)
diff --git a/code/model.py b/code/model.py
index 8109a25..4ebd957 100644
--- a/code/model.py
+++ b/code/model.py
@@ -2,6 +2,7 @@ import torch.nn as nn
from torch.nn import functional as F
import torch
import pytorch_lightning as pl
+from rmi import RMILoss
from torchvision.utils import make_grid
import segmentation_models_pytorch as smp
@@ -29,12 +30,12 @@ class TestModule(pl.LightningModule):
# Define the model
self.model = smp.Unet(
- encoder_name="densenet121", # Also consider using smaller or larger encoders
+ encoder_name="resnet34", # Also consider using smaller or larger encoders
encoder_weights= "imagenet", # Do the pretrained weights help? Try with or without
in_channels=1, # We use 1 chanel transmittance as input
classes=1, # classes == output channels. We use one output channel for cyto data
)
- self.loss_f = torch.nn.MSELoss()
+ self.loss_f = RMILoss(with_logits=True) #torch.nn.MSELoss()
def forward(self, x):
x = self.model(x)
--
GitLab