Spaces:
Sleeping
Sleeping
Upload app.py
Browse files
app.py
ADDED
@@ -0,0 +1,87 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import json
|
3 |
+
import torch
|
4 |
+
from PIL import Image
|
5 |
+
from torchvision import transforms
|
6 |
+
import matplotlib.pyplot as plt
|
7 |
+
import gradio as gr
|
8 |
+
from io import BytesIO
|
9 |
+
|
10 |
+
from vit_model import vit_base_patch16_224_in21k as create_model
|
11 |
+
|
12 |
+
def classify_image(img):
|
13 |
+
# Your existing code here, modified to use `img_path` as input
|
14 |
+
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
15 |
+
|
16 |
+
data_transform = transforms.Compose(
|
17 |
+
[transforms.Resize(256),
|
18 |
+
transforms.CenterCrop(224),
|
19 |
+
transforms.ToTensor(),
|
20 |
+
transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5])])
|
21 |
+
|
22 |
+
# [N, C, H, W]
|
23 |
+
img = data_transform(img)
|
24 |
+
# expand batch dimension
|
25 |
+
img = torch.unsqueeze(img, dim=0)
|
26 |
+
|
27 |
+
# read class_indict
|
28 |
+
json_path = 'F:\mushroom_project\VIT\class_indices.json'
|
29 |
+
assert os.path.exists(json_path), "file: '{}' dose not exist.".format(json_path)
|
30 |
+
|
31 |
+
with open(json_path, "r") as f:
|
32 |
+
class_indict = json.load(f)
|
33 |
+
|
34 |
+
# create model
|
35 |
+
model = create_model(num_classes=370, has_logits=False).to(device)
|
36 |
+
# load model weights
|
37 |
+
model_weight_path = "F:\mushroom_project\VIT\pretrain_30_weights\\best_model.pth"
|
38 |
+
#load no pretrain model path
|
39 |
+
#model_weight_path = "F:\mushroom_project\VIT\no_pretrain_weights\best_model.pth"
|
40 |
+
model.load_state_dict(torch.load(model_weight_path, map_location=device))
|
41 |
+
model.eval()
|
42 |
+
with torch.no_grad():
|
43 |
+
# predict class
|
44 |
+
output = torch.squeeze(model(img.to(device))).cpu()
|
45 |
+
predict = torch.softmax(output, dim=0)
|
46 |
+
predict_cla = torch.argmax(predict).numpy()
|
47 |
+
|
48 |
+
print_res = "class: {} prob: {:.3}".format(class_indict[str(predict_cla)],
|
49 |
+
predict[predict_cla].numpy())
|
50 |
+
|
51 |
+
# Combine the two lists into a list of tuples
|
52 |
+
combined_list = list(zip(class_indict, predict))
|
53 |
+
|
54 |
+
# Sort the combined list by the 'predict' values in descending order
|
55 |
+
sorted_combined_list = sorted(combined_list, key=lambda x: x[1], reverse=True)
|
56 |
+
|
57 |
+
# Determine the position you are currently interested in
|
58 |
+
current_position = 5 # Example position
|
59 |
+
|
60 |
+
# Get the previous five elements from the sorted list
|
61 |
+
# Ensure that the index does not go below zero
|
62 |
+
start_index = max(current_position - 5, 0)
|
63 |
+
previous_five = sorted_combined_list[start_index:current_position]
|
64 |
+
|
65 |
+
joined_string = ""
|
66 |
+
for i in previous_five:
|
67 |
+
#print("class: {:10} prob: {:.3}".format(class_indict[str(i[0])], i[1].numpy()))
|
68 |
+
joined_string += ("class: {:10} prob: {:.3}".format(class_indict[str(i[0])], i[1].numpy())) + "\n"
|
69 |
+
|
70 |
+
#print(joined_string)
|
71 |
+
plt.title(joined_string)
|
72 |
+
plt.tight_layout()
|
73 |
+
fig = plt.figure()
|
74 |
+
return joined_string
|
75 |
+
|
76 |
+
# Create a Gradio interface
|
77 |
+
iface = gr.Interface(
|
78 |
+
fn=classify_image,
|
79 |
+
inputs=gr.Image(type='pil'),
|
80 |
+
outputs=gr.Textbox(),
|
81 |
+
title="Mushrrom Image Classification",
|
82 |
+
description="Upload a mushroom image to classify."
|
83 |
+
)
|
84 |
+
|
85 |
+
# Run the Gradio app
|
86 |
+
#if __name__ == '__main__':
|
87 |
+
iface.launch()
|