Add weights to git lfs

.gitattributes

@@ -0,0 +1 @@
+models/weights.h5 filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -3,6 +3,5 @@ __pycache__/
 .DS_Store
 **/*.png
 **/*.npy
-**/*.h5
 **/*.nii.gz
 /data/raw

models/weights.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:069f52eac88d5a6fe114b27ce6cbd9868c7432d54ef63bc6a502355058bd2641
+size 93275616

notebooks/train.ipynb

@@ -8,7 +8,6 @@
    "source": [
     "import os\n",
     "from skimage.transform import resize\n",
-    "from skimage.io import imsave\n",
     "import numpy as np\n",
     "from skimage.segmentation import mark_boundaries\n",
     "from keras.models import Model\n",
@@ -17,9 +16,8 @@
     "from keras.callbacks import ModelCheckpoint\n",
     "from keras import backend as K\n",
     "from skimage.exposure import rescale_intensity\n",
-    "from keras.callbacks import History\n",
     "from skimage import io\n",
-    "from data import load_train_data, load_test_data"
+    "from ..src.data.data_loader import load_train_data, load_test_data"
    ]
   },
   {

src/utils/upload_dataset.py

@@ -0,0 +1,36 @@
+from datasets import Dataset, Features, Value
+import os
+import nibabel as nib
+
+# Define the paths to your raw data files
+raw_data_path = './data/raw'
+
+# Function to load NIfTI files
+def load_nifti(file_path):
+    nifti = nib.load(file_path)
+    return nifti.get_fdata()
+
+# Create a list to hold the data
+data = []
+
+# Iterate over the files in the raw data directory
+for file_name in os.listdir(raw_data_path):
+    if file_name.endswith('.nii.gz'):
+        file_path = os.path.join(raw_data_path, file_name)
+        data.append({
+            'file_name': file_name,
+            'data': load_nifti(file_path).tolist()  # Convert to list for serialization
+        })
+
+# Define the features of the dataset
+features = Features({
+    'file_name': Value('string'),
+    'data': Value('float32', id='data')
+})
+
+# Create Dataset object
+dataset = Dataset.from_dict({'file_name': [d['file_name'] for d in data], 'data': [d['data'] for d in data]}, features=features)
+
+if __name__ == "__main__":
+    # Push the dataset to Hugging Face
+    dataset.push_to_hub("ashutosh-pathak/liver-segmentation")