File size: 6,902 Bytes
09823ea
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
62cebeb
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
# app.py

import gradio as gr
import torch
import numpy as np
import matplotlib.pyplot as plt
import cv2
import os
import sys
import json
from PIL import Image

# Add project root to sys.path
current_dir = os.path.dirname(os.path.abspath(__file__))
project_root = os.path.dirname(current_dir)
if project_root not in sys.path:
    sys.path.append(project_root)

# Internal modules
from src.model.architecture import build_model
from src.model.gradcam import GradCAMPlusPlus as GradCAM
from src.utils.config import BEST_MODEL_PATH

# Load disease information
with open("disease_info.json", "r") as f:
    DISEASE_INFO = json.load(f)

# Load label mapping
with open("models/labels.json", "r") as f:
    idx_to_class = json.load(f)

# Load model
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
num_classes = len(idx_to_class)

model = build_model(num_classes=num_classes, freeze_backbone=False)
model.load_state_dict(torch.load(BEST_MODEL_PATH, map_location=device))
model = model.to(device)
model.eval()

target_layer = model.features[-1]

# Sample images
SAMPLE_DIR = "sample_images"
sample_choices = sorted(os.listdir(SAMPLE_DIR))

# Utility Functions
def beautify_name(raw_classname):
    parts = raw_classname.split("___")
    if len(parts) >= 3:
        plant = parts[1].title()
        disease = parts[2].replace("_", " ").replace("(", "").replace(")", "").replace("__", " ").title()
        return plant, disease
    else:
        return "Unknown", "Unknown"

def generate_gradcam(model, input_tensor, target_layer):
    gradcam = GradCAM(model, target_layer)
    cam = gradcam.generate(input_tensor)
    return cam

def preprocess_image(image):
    from torchvision import transforms
    transform = transforms.Compose([
        transforms.Resize((224, 224)),
        transforms.ToTensor(),
        transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
    ])
    return transform(image)

def predict(image):
    if image is None:
        return None, None, None, None, None

    model.eval()
    image_tensor = preprocess_image(image).unsqueeze(0).to(device)

    with torch.no_grad():
        output = model(image_tensor)
        probs = torch.softmax(output, dim=1).cpu().numpy()[0]

    top3_indices = probs.argsort()[-3:][::-1]
    top3_classes = [(idx_to_class[str(idx)], float(probs[idx])) for idx in top3_indices]

    pred_class, pred_prob = top3_classes[0]
    plant, disease = beautify_name(pred_class)

    # Format Top-3 nicely
    top3_text = ""
    for i, (c, p) in enumerate(top3_classes, 1):
        c_plant, c_disease = beautify_name(c)
        top3_text += f"{i}. {c_plant} - {c_disease} ({p*100:.2f}%)\n"

    # GradCAM
    cam = generate_gradcam(model, image_tensor, target_layer)

    img_np = np.array(image) / 255.0

    # 🔥 Resize uploaded image to 224x224 for matching heatmap
    img_np_resized = cv2.resize(img_np, (224, 224))

    img_gray = cv2.cvtColor(np.uint8(img_np_resized * 255), cv2.COLOR_RGB2GRAY) / 255.0
    img_gray_3ch = np.stack([img_gray]*3, axis=-1)

    heatmap = cv2.applyColorMap(np.uint8(255 * cam), cv2.COLORMAP_PLASMA)
    heatmap = np.float32(heatmap) / 255

    overlay = heatmap + img_gray_3ch
    overlay = overlay / np.max(overlay)

    fig, ax = plt.subplots(figsize=(5,5))
    ax.imshow(overlay)
    ax.axis("off")
    cbar = plt.colorbar(plt.cm.ScalarMappable(cmap='plasma'), orientation='horizontal', pad=0.05, ax=ax)
    cbar.set_ticks([0, 0.5, 1])
    cbar.set_ticklabels(['Low Focus', 'Medium Focus', 'High Focus'])
    plt.tight_layout()

    cam_path = "cam_output.png"
    fig.savefig(cam_path)
    plt.close(fig)

    # Health Status + Disease Info
    if "healthy" in pred_class.lower():
        health_status = f"{plant} - Healthy"
        identified_disease = "None"
        disease_info_text = "✅ No disease detected."
    else:
        health_status = f"{plant} - Diseased"
        identified_disease = disease
        disease_data = DISEASE_INFO.get(pred_class, {})
        disease_info_text = f"""
**Symptoms:** {disease_data.get('symptoms', 'No information available.')}

**Causes:** {disease_data.get('causes', 'No information available.')}

**Disease Cycle:** {disease_data.get('disease_cycle', 'No information available.')}

**Care & Treatment:** {disease_data.get('care_treatment', 'No information available.')}

[Learn more on Wikipedia]({disease_data.get('wiki_url', '#')})
"""

    alert = None
    if pred_prob < 0.6:
        alert = "⚠️ Low confidence in prediction! Please verify manually."

    return health_status, identified_disease, top3_text, alert, cam_path, disease_info_text

def load_sample_image(sample_name):
    img_path = os.path.join(SAMPLE_DIR, sample_name)
    img = Image.open(img_path).convert("RGB")
    return img

# Interface
title = "CropGuard: Leaf Disease Detector"
copyright_text = "© 2025 Made by [Arka Mitra](https://github.com/mitraarka27)"
instruction_text = """
Upload a clear image of a **potato**, **tomato**, or **grape** leaf.

CropGuard will predict:
- Whether the leaf is **healthy** or **diseased**.
- The likely disease (if any).
- Where the model focused its attention.

⚡ **Note**: Currently supports only **Potato, Tomato, Grape** leaves.
"""

with gr.Blocks(theme="default") as app:
    with gr.Row():
        gr.Markdown(f"<h1 style='text-align: center;'>{title}</h1>")
    gr.Markdown("<p style='text-align: center;'>© 2025 Made by <a href='https://github.com/mitraarka27' target='_blank'>Arka Mitra</a></p>")
    with gr.Row():
        with gr.Column(scale=2):
            upload = gr.Image(
                type="pil",
                sources=["upload", "webcam", "clipboard"],
                label="Upload, Capture, or Paste Leaf Image"
            )
            gr.Markdown("**OR** choose from sample images below:")
            sample_dropdown = gr.Dropdown(choices=sample_choices, label="Select a Sample Image")
            load_btn = gr.Button("Load Sample Image")
            predict_btn = gr.Button("Predict", variant="primary")
            gr.Markdown(instruction_text)
            alert_box = gr.Textbox(label="Prediction Alert", lines=2, interactive=False)
            top3_preds = gr.Textbox(label="Top-3 Predictions", lines=5, interactive=False)
        with gr.Column(scale=3):
            health_status = gr.Label(label="Plant Health Status")
            disease_name = gr.Label(label="Identified Disease (includes details)")
            disease_info = gr.Markdown()
            heatmap = gr.Image(label="Model Focus Heatmap")

    load_btn.click(
        fn=load_sample_image,
        inputs=[sample_dropdown],
        outputs=[upload]
    )

    predict_btn.click(
        fn=predict,
        inputs=[upload],
        outputs=[health_status, disease_name, top3_preds, alert_box, heatmap, disease_info]
    )

if __name__ == "__main__":
    app.launch(server_name="0.0.0.0", server_port=7860)