Spaces:
Runtime error
Runtime error
add implementation/interface
Browse files
.DS_Store
ADDED
Binary file (6.15 kB). View file
|
|
app.py
CHANGED
@@ -2,21 +2,22 @@ import datasets
|
|
2 |
import gradio as gr
|
3 |
from transformers import AutoFeatureExtractor, AutoModelForImageClassification
|
4 |
|
5 |
-
def classify(im):
|
6 |
-
features = feature_extractor(im, return_tensors='pt')
|
7 |
-
logits = model(features['pixel_values'])[-1]
|
8 |
-
probability = torch.nn.functional.softmax(logits, dim=-1)
|
9 |
-
probs = probability[0].detach().numpy()
|
10 |
-
confidences = {label: float(probs[i]) for i, label in enumerate(labels)}
|
11 |
-
return confidences
|
12 |
-
|
13 |
dataset = datasets.load_dataset('beans', 'full_size')
|
14 |
|
|
|
15 |
extractor = AutoFeatureExtractor.from_pretrained('saved_model_files')
|
16 |
model = AutoModelForImageClassification.from_pretrained('saved_model_files')
|
17 |
|
18 |
labels = dataset['train'].features['labels'].names
|
19 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
20 |
interface = gr.Interface(fn=classify, inputs=gr.Image(shape=(200, 200)), outputs=gr.outputs.Label(num_top_classes=3),
|
21 |
examples=['leaf1.png', 'leaf2.png'], title='Leaf Classification App', description='Check if the leaves of your plant are healthy!', flagging_dir='flagged_examples/')
|
22 |
|
|
|
2 |
import gradio as gr
|
3 |
from transformers import AutoFeatureExtractor, AutoModelForImageClassification
|
4 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
dataset = datasets.load_dataset('beans', 'full_size')
|
6 |
|
7 |
+
feature_extractor = ViTFeatureExtractor.from_pretrained('google/vit-base-patch16-224')
|
8 |
extractor = AutoFeatureExtractor.from_pretrained('saved_model_files')
|
9 |
model = AutoModelForImageClassification.from_pretrained('saved_model_files')
|
10 |
|
11 |
labels = dataset['train'].features['labels'].names
|
12 |
|
13 |
+
def classify(im):
|
14 |
+
features = extractor(im, return_tensors='pt')
|
15 |
+
logits = model(features['pixel_values'])[-1]
|
16 |
+
probability = torch.nn.functional.softmax(logits, dim=-1)
|
17 |
+
probs = probability[0].detach().numpy()
|
18 |
+
confidences = {label: float(probs[i]) for i, label in enumerate(labels)}
|
19 |
+
return confidences
|
20 |
+
|
21 |
interface = gr.Interface(fn=classify, inputs=gr.Image(shape=(200, 200)), outputs=gr.outputs.Label(num_top_classes=3),
|
22 |
examples=['leaf1.png', 'leaf2.png'], title='Leaf Classification App', description='Check if the leaves of your plant are healthy!', flagging_dir='flagged_examples/')
|
23 |
|