Spaces:
Sleeping
Sleeping
debugging: xai output distortion
Browse files- app.py +2 -2
- explanations.py +2 -6
app.py
CHANGED
@@ -141,7 +141,7 @@ def classify_image(input_image, model_name):
|
|
141 |
from inference_resnet import inference_resnet_finer
|
142 |
model,n_classes= get_model(model_name)
|
143 |
result = inference_resnet_finer(input_image,model,size=600,n_classes=n_classes)
|
144 |
-
return result
|
145 |
elif 'Mummified 170' ==model_name:
|
146 |
from inference_resnet import inference_resnet_finer
|
147 |
model, n_classes= get_model(model_name)
|
@@ -169,7 +169,7 @@ def get_embeddings(input_image,model_name):
|
|
169 |
from inference_resnet import inference_resnet_embedding
|
170 |
model,n_classes= get_model(model_name)
|
171 |
result = inference_resnet_embedding(input_image,model,size=600,n_classes=n_classes)
|
172 |
-
return result
|
173 |
elif 'Mummified 170' ==model_name:
|
174 |
from inference_resnet import inference_resnet_embedding
|
175 |
model, n_classes= get_model(model_name)
|
|
|
141 |
from inference_resnet import inference_resnet_finer
|
142 |
model,n_classes= get_model(model_name)
|
143 |
result = inference_resnet_finer(input_image,model,size=600,n_classes=n_classes)
|
144 |
+
return result
|
145 |
elif 'Mummified 170' ==model_name:
|
146 |
from inference_resnet import inference_resnet_finer
|
147 |
model, n_classes= get_model(model_name)
|
|
|
169 |
from inference_resnet import inference_resnet_embedding
|
170 |
model,n_classes= get_model(model_name)
|
171 |
result = inference_resnet_embedding(input_image,model,size=600,n_classes=n_classes)
|
172 |
+
return result
|
173 |
elif 'Mummified 170' ==model_name:
|
174 |
from inference_resnet import inference_resnet_embedding
|
175 |
model, n_classes= get_model(model_name)
|
explanations.py
CHANGED
@@ -29,13 +29,8 @@ def preprocess_image(image, output_size=(300, 300)):
|
|
29 |
return image_resized
|
30 |
|
31 |
def show(img, output_size,p=False, **kwargs):
|
32 |
-
img = np.array(img, dtype=np.float32)
|
33 |
-
h, w = img.shape[:2]
|
34 |
-
print(h,w)
|
35 |
|
36 |
-
img = preprocess_image(img, output_size=(output_size,output_size))
|
37 |
-
h, w = img.shape[:2]
|
38 |
-
print(h,w)
|
39 |
|
40 |
# check if channel first
|
41 |
if img.shape[0] == 1:
|
@@ -53,6 +48,7 @@ def show(img, output_size,p=False, **kwargs):
|
|
53 |
# check if clip percentile
|
54 |
if p is not False:
|
55 |
img = np.clip(img, np.percentile(img, p), np.percentile(img, 100-p))
|
|
|
56 |
plt.imshow(img, **kwargs)
|
57 |
plt.axis('off')
|
58 |
|
|
|
29 |
return image_resized
|
30 |
|
31 |
def show(img, output_size,p=False, **kwargs):
|
|
|
|
|
|
|
32 |
|
33 |
+
#img = preprocess_image(img, output_size=(output_size,output_size))
|
|
|
|
|
34 |
|
35 |
# check if channel first
|
36 |
if img.shape[0] == 1:
|
|
|
48 |
# check if clip percentile
|
49 |
if p is not False:
|
50 |
img = np.clip(img, np.percentile(img, p), np.percentile(img, 100-p))
|
51 |
+
img = preprocess_image(img, output_size=(output_size,output_size))
|
52 |
plt.imshow(img, **kwargs)
|
53 |
plt.axis('off')
|
54 |
|