Spaces:
Sleeping
Sleeping
keremoktay1
commited on
Commit
•
97eac4c
1
Parent(s):
61c039d
Update app.py
Browse files
app.py
CHANGED
@@ -3,9 +3,6 @@ import pickle
|
|
3 |
import numpy as np
|
4 |
from tensorflow.keras.models import load_model
|
5 |
from PIL import Image
|
6 |
-
import torch
|
7 |
-
import torchvision.transforms as transforms
|
8 |
-
import torchvision.models as models
|
9 |
|
10 |
# Update the path to the directory where your models are stored
|
11 |
model_directory = '/Users/keremoktay/Desktop/Gradio/'
|
@@ -13,8 +10,8 @@ model_directory = '/Users/keremoktay/Desktop/Gradio/'
|
|
13 |
# Load the decision tree and KNN models
|
14 |
dt_models = {
|
15 |
"2": pickle.load(open('dt2.pkl', 'rb')),
|
16 |
-
"3": pickle.load(open('dt3.pkl', 'rb')),
|
17 |
-
"5": pickle.load(open('dt5.pkl', 'rb'))
|
18 |
}
|
19 |
knn_models = {
|
20 |
"1": pickle.load(open('knn1.pkl', 'rb')),
|
@@ -29,38 +26,27 @@ nn_models = {
|
|
29 |
"3": load_model('cnn_3layer.h5')
|
30 |
}
|
31 |
|
32 |
-
|
33 |
-
|
34 |
-
model - models.mobilenet_v2()
|
35 |
-
model.load_state_dict(torch.load('model_weights.pth'))
|
36 |
-
model.eval()
|
37 |
-
|
38 |
def preprocess_image_for_ml(image_path):
|
39 |
-
|
40 |
-
|
41 |
-
|
|
|
|
|
|
|
|
|
|
|
42 |
|
43 |
def preprocess_image_for_cnn(image_path, target_size=(128, 128)):
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
transforms.ToTensor(),
|
55 |
-
transforms.Normalize(mean=[0.485,0.456,0.406], std=[0.229,0.224,0.225])
|
56 |
-
])
|
57 |
-
img = Image.open(image_path)
|
58 |
-
img_t = preprocess(img)
|
59 |
-
batch_t = torch.unsqueeze(img_t, 0)
|
60 |
-
return batch_t
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
|
65 |
class_names = {
|
66 |
0: "Angular Leaf Spot",
|
@@ -69,32 +55,37 @@ class_names = {
|
|
69 |
}
|
70 |
|
71 |
def classify_image_with_decision_tree(image, depth):
|
72 |
-
|
73 |
-
|
74 |
-
|
75 |
-
|
|
|
|
|
|
|
|
|
|
|
76 |
|
77 |
def classify_image_with_knn(image, k):
|
78 |
-
|
79 |
-
|
80 |
-
|
81 |
-
|
|
|
|
|
|
|
|
|
|
|
82 |
|
83 |
def classify_image_with_neural_network(image, layers):
|
84 |
-
|
85 |
-
|
86 |
-
|
87 |
-
|
88 |
-
|
89 |
-
|
90 |
-
|
91 |
-
|
92 |
-
|
93 |
-
_, indices = torch.topk(out, 1)
|
94 |
-
|
95 |
-
return indices[0]
|
96 |
-
|
97 |
-
|
98 |
|
99 |
# Create Gradio interface
|
100 |
with gr.Blocks() as demo:
|
@@ -124,11 +115,4 @@ with gr.Blocks() as demo:
|
|
124 |
classify_button_nn = gr.Button("Classify with Neural Network")
|
125 |
classify_button_nn.click(classify_image_with_neural_network, inputs=[image_input_nn, dropdown_layers], outputs=output_nn)
|
126 |
|
127 |
-
with gr.Tab("mobilenet Model"):
|
128 |
-
with gr.Row():
|
129 |
-
image_input_pytorch = gr.Image(type="filepath")
|
130 |
-
output_pytorch = gr.Textbox(label="ResNet-18 Model Output")
|
131 |
-
classify_button_pytorch = gr.Button("Classify with ResNet-18 Model")
|
132 |
-
classify_button_pytorch.click(classify_image_with_pytorch, inputs=[image_input_pytorch], outputs=output_pytorch)
|
133 |
-
|
134 |
demo.launch()
|
|
|
3 |
import numpy as np
|
4 |
from tensorflow.keras.models import load_model
|
5 |
from PIL import Image
|
|
|
|
|
|
|
6 |
|
7 |
# Update the path to the directory where your models are stored
|
8 |
model_directory = '/Users/keremoktay/Desktop/Gradio/'
|
|
|
10 |
# Load the decision tree and KNN models
|
11 |
dt_models = {
|
12 |
"2": pickle.load(open('dt2.pkl', 'rb')),
|
13 |
+
"3": pickle.load(open('dt3.pkl', 'rb')),
|
14 |
+
"5": pickle.load(open('dt5.pkl', 'rb'))
|
15 |
}
|
16 |
knn_models = {
|
17 |
"1": pickle.load(open('knn1.pkl', 'rb')),
|
|
|
26 |
"3": load_model('cnn_3layer.h5')
|
27 |
}
|
28 |
|
|
|
|
|
|
|
|
|
|
|
|
|
29 |
def preprocess_image_for_ml(image_path):
|
30 |
+
try:
|
31 |
+
# For traditional ML models, just flatten the image without resizing
|
32 |
+
image = Image.open(image_path)
|
33 |
+
image_array = np.asarray(image).flatten().reshape(1, -1)
|
34 |
+
return image_array
|
35 |
+
except Exception as e:
|
36 |
+
print(f"Error in preprocess_image_for_ml: {e}")
|
37 |
+
raise e
|
38 |
|
39 |
def preprocess_image_for_cnn(image_path, target_size=(128, 128)):
|
40 |
+
try:
|
41 |
+
# For CNN models, convert to RGB, resize to 128x128, normalize, and add a batch dimension
|
42 |
+
image = Image.open(image_path).convert('RGB')
|
43 |
+
image = image.resize(target_size)
|
44 |
+
image_array = np.asarray(image) / 255.0
|
45 |
+
image_array = np.expand_dims(image_array, axis=0)
|
46 |
+
return image_array
|
47 |
+
except Exception as e:
|
48 |
+
print(f"Error in preprocess_image_for_cnn: {e}")
|
49 |
+
raise e
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
50 |
|
51 |
class_names = {
|
52 |
0: "Angular Leaf Spot",
|
|
|
55 |
}
|
56 |
|
57 |
def classify_image_with_decision_tree(image, depth):
|
58 |
+
try:
|
59 |
+
image_array = preprocess_image_for_ml(image)
|
60 |
+
model = dt_models.get(depth)
|
61 |
+
prediction = model.predict(image_array)
|
62 |
+
# Sayısal tahmini hastalık ismine dönüştür
|
63 |
+
return class_names[int(prediction)]
|
64 |
+
except Exception as e:
|
65 |
+
print(f"Error in classify_image_with_decision_tree: {e}")
|
66 |
+
raise e
|
67 |
|
68 |
def classify_image_with_knn(image, k):
|
69 |
+
try:
|
70 |
+
image_array = preprocess_image_for_ml(image)
|
71 |
+
model = knn_models.get(k)
|
72 |
+
prediction = model.predict(image_array)
|
73 |
+
# Sayısal tahmini hastalık ismine dönüştür
|
74 |
+
return class_names[int(prediction)]
|
75 |
+
except Exception as e:
|
76 |
+
print(f"Error in classify_image_with_knn: {e}")
|
77 |
+
raise e
|
78 |
|
79 |
def classify_image_with_neural_network(image, layers):
|
80 |
+
try:
|
81 |
+
image_array = preprocess_image_for_cnn(image)
|
82 |
+
model = nn_models.get(layers)
|
83 |
+
prediction = model.predict(image_array)
|
84 |
+
# En yüksek tahmin değerine sahip indeksi bul ve hastalık ismine dönüştür
|
85 |
+
return class_names[np.argmax(prediction, axis=1)[0]]
|
86 |
+
except Exception as e:
|
87 |
+
print(f"Error in classify_image_with_neural_network: {e}")
|
88 |
+
raise e
|
|
|
|
|
|
|
|
|
|
|
89 |
|
90 |
# Create Gradio interface
|
91 |
with gr.Blocks() as demo:
|
|
|
115 |
classify_button_nn = gr.Button("Classify with Neural Network")
|
116 |
classify_button_nn.click(classify_image_with_neural_network, inputs=[image_input_nn, dropdown_layers], outputs=output_nn)
|
117 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
118 |
demo.launch()
|