akhaliq HF staff commited on
Commit
ea25a98
1 Parent(s): 9c05d1d

Create app.py

Browse files
Files changed (1) hide show
  1. app.py +53 -0
app.py ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import torch
2
+ from PIL import Image
3
+ from torchvision import transforms
4
+ import gradio as gr
5
+
6
+ model = torch.hub.load('pytorch/vision:v0.9.0', 'googlenet', pretrained=True)
7
+ model.eval()
8
+
9
+ torch.hub.download_url_to_file("https://github.com/pytorch/hub/raw/master/images/dog.jpg", "dog.jpg")
10
+
11
+ # sample execution (requires torchvision)
12
+ def inference(input_image):
13
+ preprocess = transforms.Compose([
14
+ transforms.Resize(256),
15
+ transforms.CenterCrop(224),
16
+ transforms.ToTensor(),
17
+ transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
18
+ ])
19
+ input_tensor = preprocess(input_image)
20
+ input_batch = input_tensor.unsqueeze(0) # create a mini-batch as expected by the model
21
+
22
+ # move the input and model to GPU for speed if available
23
+ if torch.cuda.is_available():
24
+ input_batch = input_batch.to('cuda')
25
+ model.to('cuda')
26
+
27
+ with torch.no_grad():
28
+ output = model(input_batch)
29
+ # The output has unnormalized scores. To get probabilities, you can run a softmax on it.
30
+ probabilities = torch.nn.functional.softmax(output[0], dim=0)
31
+ # Download ImageNet labels
32
+ !wget https://raw.githubusercontent.com/pytorch/hub/master/imagenet_classes.txt
33
+ # Read the categories
34
+ with open("imagenet_classes.txt", "r") as f:
35
+ categories = [s.strip() for s in f.readlines()]
36
+ # Show top categories per image
37
+ top5_prob, top5_catid = torch.topk(probabilities, 5)
38
+ result = {}
39
+ for i in range(top5_prob.size(0)):
40
+ result[categories[top5_catid[i]]] = top5_prob[i].item()
41
+ return result
42
+
43
+ inputs = gr.inputs.Image(type='pil')
44
+ outputs = gr.outputs.Label(type="confidences",num_top_classes=5)
45
+
46
+ title = "GOOGLENET"
47
+ description = "Gradio demo for GOOGLENET, GoogLeNet was based on a deep convolutional neural network architecture codenamed Inception which won ImageNet 2014. To use it, simply upload your image, or click one of the examples to load them. Read more at the links below."
48
+ article = "<p style='text-align: center'><a href='https://arxiv.org/abs/1409.4842'>Going Deeper with Convolutions</a> | <a href='https://github.com/pytorch/vision/blob/master/torchvision/models/googlenet.py'>Github Repo</a></p>"
49
+
50
+ examples = [
51
+ ['dog.jpg']
52
+ ]
53
+ gr.Interface(inference, inputs, outputs, title=title, description=description, article=article, examples=examples, analytics_enabled=False).launch()