Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -3,6 +3,7 @@ import os
|
|
3 |
import torch
|
4 |
from model import create_effnetb2_model
|
5 |
from timeit import default_timer as timer
|
|
|
6 |
from typing import Tuple, Dict
|
7 |
|
8 |
class_names = ['pizza', 'steak', 'sushi']
|
@@ -14,14 +15,16 @@ effnetb2.load_state_dict(
|
|
14 |
)
|
15 |
)
|
16 |
|
17 |
-
|
|
|
|
|
18 |
# Start a timer
|
19 |
start_time = timer()
|
20 |
-
# Transform the input image for use
|
21 |
img = effnetb2_transforms(img).unsqueeze(0)
|
22 |
# Put model into eval mode, make prediction
|
23 |
effnetb2.eval()
|
24 |
-
with torch.
|
25 |
pred_probs = torch.softmax(effnetb2(img), dim=1)
|
26 |
# Create a prediction labal and prediction probability dictionary
|
27 |
pred_labels_and_probs = {class_names[i]:float(pred_probs[0][i]) for i in range(len(class_names))}
|
@@ -32,20 +35,26 @@ def predict(img):
|
|
32 |
return pred_labels_and_probs, pred_time
|
33 |
|
34 |
|
35 |
-
|
|
|
|
|
|
|
36 |
description = 'An [EfficientNetB2 feature extractor](https://pytorch.org/vision/main/models/generated/torchvision.models.efficientnet_b2.html)'
|
37 |
-
article = 'Created with Pytorch model deployment'
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
|
|
|
|
|
|
|
3 |
import torch
|
4 |
from model import create_effnetb2_model
|
5 |
from timeit import default_timer as timer
|
6 |
+
from PIL import Image
|
7 |
from typing import Tuple, Dict
|
8 |
|
9 |
class_names = ['pizza', 'steak', 'sushi']
|
|
|
15 |
)
|
16 |
)
|
17 |
|
18 |
+
### Prediction function: EffNetB2 ###
|
19 |
+
|
20 |
+
def predict(img: Image.Image) -> Tuple[Dict[str, float], float]:
|
21 |
# Start a timer
|
22 |
start_time = timer()
|
23 |
+
# Transform the input image for use with EffNetB2
|
24 |
img = effnetb2_transforms(img).unsqueeze(0)
|
25 |
# Put model into eval mode, make prediction
|
26 |
effnetb2.eval()
|
27 |
+
with torch.no_grad():
|
28 |
pred_probs = torch.softmax(effnetb2(img), dim=1)
|
29 |
# Create a prediction labal and prediction probability dictionary
|
30 |
pred_labels_and_probs = {class_names[i]:float(pred_probs[0][i]) for i in range(len(class_names))}
|
|
|
35 |
return pred_labels_and_probs, pred_time
|
36 |
|
37 |
|
38 |
+
### Gradio app ###
|
39 |
+
|
40 |
+
# Create title, description and article strings
|
41 |
+
title = "FoodVision Mini ππ₯©π£"
|
42 |
description = 'An [EfficientNetB2 feature extractor](https://pytorch.org/vision/main/models/generated/torchvision.models.efficientnet_b2.html)'
|
43 |
+
article = 'Created by Arik Kodenzov with Pytorch model deployment'
|
44 |
+
|
45 |
+
# Create examples list from "examples/" directory
|
46 |
+
example_list = [["examples/" + example] for example in os.listdir("examples")]
|
47 |
+
|
48 |
+
# Create the Gradio demo
|
49 |
+
demo = gr.Interface(fn=predict, # mapping function from input to output
|
50 |
+
inputs=gr.Image(type="pil"), # what are the inputs?
|
51 |
+
outputs=[gr.Label(num_top_classes=3, label="Predictions"), # what are the outputs?
|
52 |
+
gr.Number(label="Prediction time (s)")], # our fn has two outputs, therefore we have two outputs
|
53 |
+
# Create examples list from "examples/" directory
|
54 |
+
examples=example_list,
|
55 |
+
title=title,
|
56 |
+
description=description,
|
57 |
+
article=article)
|
58 |
+
|
59 |
+
# Launch the demo!
|
60 |
+
demo.launch()
|