Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -1,7 +1,7 @@
|
|
1 |
import cv2
|
2 |
-
import torch
|
3 |
import numpy as np
|
4 |
from PIL import Image
|
|
|
5 |
from torchvision import models, transforms
|
6 |
from ultralytics import YOLO
|
7 |
import gradio as gr
|
@@ -11,10 +11,8 @@ import torch.nn as nn
|
|
11 |
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
|
12 |
|
13 |
# Load models
|
14 |
-
yolo_model = YOLO('best.pt') # Make sure this file is uploaded
|
15 |
resnet = models.resnet50(pretrained=False)
|
16 |
-
|
17 |
-
# Modify ResNet for 3 classes
|
18 |
resnet.fc = nn.Linear(resnet.fc.in_features, 3)
|
19 |
resnet.load_state_dict(torch.load('rice_resnet_model.pth', map_location=device))
|
20 |
resnet = resnet.to(device)
|
@@ -38,12 +36,17 @@ def classify_crop(crop_img):
|
|
38 |
_, predicted = torch.max(output, 1)
|
39 |
return class_labels[predicted.item()]
|
40 |
|
41 |
-
def detect_and_classify(
|
42 |
-
"""Process
|
43 |
-
|
|
|
|
|
|
|
|
|
44 |
results = yolo_model(image)[0]
|
45 |
boxes = results.boxes.xyxy.cpu().numpy()
|
46 |
-
|
|
|
47 |
for box in boxes:
|
48 |
x1, y1, x2, y2 = map(int, box[:4])
|
49 |
crop = image[y1:y2, x1:x2]
|
@@ -52,35 +55,36 @@ def detect_and_classify(image):
|
|
52 |
|
53 |
# Draw bounding box and label
|
54 |
cv2.rectangle(image, (x1, y1), (x2, y2), (0, 255, 0), 2)
|
55 |
-
cv2.putText(image,
|
56 |
-
|
|
|
|
|
|
|
|
|
|
|
57 |
|
|
|
58 |
return Image.fromarray(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
|
59 |
|
60 |
-
# Gradio
|
61 |
-
with gr.Blocks(title="
|
62 |
gr.Markdown("""
|
63 |
-
|
64 |
-
|
65 |
""")
|
66 |
|
67 |
with gr.Row():
|
68 |
-
|
69 |
-
|
|
|
|
|
|
|
70 |
|
71 |
-
submit_btn = gr.Button("تشخیص کریں")
|
72 |
submit_btn.click(
|
73 |
fn=detect_and_classify,
|
74 |
-
inputs=
|
75 |
outputs=output_image
|
76 |
)
|
77 |
-
|
78 |
-
gr.Examples(
|
79 |
-
examples=[["example1.jpg"], ["example2.jpg"]], # Add your example images
|
80 |
-
inputs=input_image,
|
81 |
-
outputs=output_image,
|
82 |
-
fn=detect_and_classify,
|
83 |
-
cache_examples=True
|
84 |
-
)
|
85 |
|
|
|
86 |
demo.launch()
|
|
|
1 |
import cv2
|
|
|
2 |
import numpy as np
|
3 |
from PIL import Image
|
4 |
+
import torch
|
5 |
from torchvision import models, transforms
|
6 |
from ultralytics import YOLO
|
7 |
import gradio as gr
|
|
|
11 |
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
|
12 |
|
13 |
# Load models
|
14 |
+
yolo_model = YOLO('best.pt') # Make sure this file is uploaded
|
15 |
resnet = models.resnet50(pretrained=False)
|
|
|
|
|
16 |
resnet.fc = nn.Linear(resnet.fc.in_features, 3)
|
17 |
resnet.load_state_dict(torch.load('rice_resnet_model.pth', map_location=device))
|
18 |
resnet = resnet.to(device)
|
|
|
36 |
_, predicted = torch.max(output, 1)
|
37 |
return class_labels[predicted.item()]
|
38 |
|
39 |
+
def detect_and_classify(input_image):
|
40 |
+
"""Process uploaded image"""
|
41 |
+
# Convert Gradio Image to OpenCV format
|
42 |
+
image = np.array(input_image)
|
43 |
+
image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
|
44 |
+
|
45 |
+
# YOLO Detection
|
46 |
results = yolo_model(image)[0]
|
47 |
boxes = results.boxes.xyxy.cpu().numpy()
|
48 |
+
|
49 |
+
# Process each detection
|
50 |
for box in boxes:
|
51 |
x1, y1, x2, y2 = map(int, box[:4])
|
52 |
crop = image[y1:y2, x1:x2]
|
|
|
55 |
|
56 |
# Draw bounding box and label
|
57 |
cv2.rectangle(image, (x1, y1), (x2, y2), (0, 255, 0), 2)
|
58 |
+
cv2.putText(image,
|
59 |
+
predicted_label,
|
60 |
+
(x1, y1-10),
|
61 |
+
cv2.FONT_HERSHEY_SIMPLEX,
|
62 |
+
0.9,
|
63 |
+
(36, 255, 12),
|
64 |
+
2)
|
65 |
|
66 |
+
# Convert back to RGB for Gradio
|
67 |
return Image.fromarray(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
|
68 |
|
69 |
+
# Create Gradio interface
|
70 |
+
with gr.Blocks(title="Rice Classification") as demo:
|
71 |
gr.Markdown("""
|
72 |
+
## 🍚 Rice Variety Classifier
|
73 |
+
Upload an image containing rice grains. The system will detect and classify each grain.
|
74 |
""")
|
75 |
|
76 |
with gr.Row():
|
77 |
+
with gr.Column():
|
78 |
+
image_input = gr.Image(type="pil", label="Upload Rice Image")
|
79 |
+
submit_btn = gr.Button("Analyze", variant="primary")
|
80 |
+
with gr.Column():
|
81 |
+
output_image = gr.Image(label="Detection Results", interactive=False)
|
82 |
|
|
|
83 |
submit_btn.click(
|
84 |
fn=detect_and_classify,
|
85 |
+
inputs=image_input,
|
86 |
outputs=output_image
|
87 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
88 |
|
89 |
+
# Launch the app
|
90 |
demo.launch()
|