Spaces:
Sleeping
Sleeping
qijie.wei
commited on
Commit
·
6735c2f
1
Parent(s):
c5f4ee2
update
Browse files- inference.py +4 -3
inference.py
CHANGED
|
@@ -28,7 +28,7 @@ class Inference(object):
|
|
| 28 |
def inference(self, image, modality):
|
| 29 |
assert modality in self.modality_mapping, "Modality '{}' not supported".format(modality)
|
| 30 |
|
| 31 |
-
image = self.load_image(image)
|
| 32 |
modality_idx = self.modality_mapping[modality]
|
| 33 |
modality_idx = torch.tensor([modality_idx])
|
| 34 |
with torch.no_grad():
|
|
@@ -36,19 +36,20 @@ class Inference(object):
|
|
| 36 |
output = output.data.cpu().numpy()[0][0]
|
| 37 |
output = sigmoid(output) * 255
|
| 38 |
output = output.astype(np.uint8)
|
|
|
|
| 39 |
return output
|
| 40 |
|
| 41 |
def load_image(self, image):
|
| 42 |
# Load the image and preprocess it
|
| 43 |
if isinstance(image, str):
|
| 44 |
image = cv2.imread(image)[:, :, [2, 1, 0]]
|
| 45 |
-
|
| 46 |
image = cv2.resize(image, (self.model_params['size_w'], self.model_params['size_h']))
|
| 47 |
image = image.astype(np.float32) / 255.0
|
| 48 |
image = np.transpose(image, (2, 0, 1))
|
| 49 |
image = np.expand_dims(image, axis=0)
|
| 50 |
image = torch.tensor(image)
|
| 51 |
-
return image
|
| 52 |
|
| 53 |
def load_model(self):
|
| 54 |
print('Loading model from {}'.format(self.model_path))
|
|
|
|
| 28 |
def inference(self, image, modality):
|
| 29 |
assert modality in self.modality_mapping, "Modality '{}' not supported".format(modality)
|
| 30 |
|
| 31 |
+
image, raw_h, raw_w = self.load_image(image)
|
| 32 |
modality_idx = self.modality_mapping[modality]
|
| 33 |
modality_idx = torch.tensor([modality_idx])
|
| 34 |
with torch.no_grad():
|
|
|
|
| 36 |
output = output.data.cpu().numpy()[0][0]
|
| 37 |
output = sigmoid(output) * 255
|
| 38 |
output = output.astype(np.uint8)
|
| 39 |
+
output = cv2.resize(output, (raw_w, raw_h))
|
| 40 |
return output
|
| 41 |
|
| 42 |
def load_image(self, image):
|
| 43 |
# Load the image and preprocess it
|
| 44 |
if isinstance(image, str):
|
| 45 |
image = cv2.imread(image)[:, :, [2, 1, 0]]
|
| 46 |
+
raw_h, raw_w = image.shape[:2]
|
| 47 |
image = cv2.resize(image, (self.model_params['size_w'], self.model_params['size_h']))
|
| 48 |
image = image.astype(np.float32) / 255.0
|
| 49 |
image = np.transpose(image, (2, 0, 1))
|
| 50 |
image = np.expand_dims(image, axis=0)
|
| 51 |
image = torch.tensor(image)
|
| 52 |
+
return image, raw_h, raw_w
|
| 53 |
|
| 54 |
def load_model(self):
|
| 55 |
print('Loading model from {}'.format(self.model_path))
|