Upload model
Browse files- modeling.py +8 -1
modeling.py
CHANGED
|
@@ -404,7 +404,9 @@ class TotalClassifierModel(PreTrainedModel):
|
|
| 404 |
return slices, [dicom_files[idx] for idx in indices]
|
| 405 |
return slices
|
| 406 |
|
| 407 |
-
def preprocess(
|
|
|
|
|
|
|
| 408 |
mode = mode.lower()
|
| 409 |
if mode == "2d":
|
| 410 |
x = cv2.resize(x, self.image_size)
|
|
@@ -414,6 +416,11 @@ class TotalClassifierModel(PreTrainedModel):
|
|
| 414 |
x = np.stack([cv2.resize(s, self.image_size) for s in x], axis=0)
|
| 415 |
if x.ndim == 3:
|
| 416 |
x = x[:, :, :, np.newaxis]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 417 |
return x
|
| 418 |
|
| 419 |
def crop_single_plane(
|
|
|
|
| 404 |
return slices, [dicom_files[idx] for idx in indices]
|
| 405 |
return slices
|
| 406 |
|
| 407 |
+
def preprocess(
|
| 408 |
+
self, x: np.ndarray, mode: str = "2d", torchify: bool = True
|
| 409 |
+
) -> np.ndarray:
|
| 410 |
mode = mode.lower()
|
| 411 |
if mode == "2d":
|
| 412 |
x = cv2.resize(x, self.image_size)
|
|
|
|
| 416 |
x = np.stack([cv2.resize(s, self.image_size) for s in x], axis=0)
|
| 417 |
if x.ndim == 3:
|
| 418 |
x = x[:, :, :, np.newaxis]
|
| 419 |
+
if torchify:
|
| 420 |
+
if x.ndim == 3:
|
| 421 |
+
x = torch.from_numpy(x).float().permute(2, 0, 1)
|
| 422 |
+
elif x.ndim == 4:
|
| 423 |
+
x = torch.from_numpy(x).float().permute(3, 0, 1, 2)
|
| 424 |
return x
|
| 425 |
|
| 426 |
def crop_single_plane(
|