Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import torch
|
2 |
+
from torchvision.models.detection import fasterrcnn_resnet50_fpn
|
3 |
+
from torchvision.models.detection.faster_rcnn import FastRCNNPredictor
|
4 |
+
from torchvision.transforms import functional as F
|
5 |
+
from PIL import Image, ImageDraw
|
6 |
+
import gradio as gr
|
7 |
+
|
8 |
+
# Label names
|
9 |
+
COCO_CLASSES = {
|
10 |
+
0: "Background",
|
11 |
+
1: "Without Mask",
|
12 |
+
2: "With Mask",
|
13 |
+
3: "Incorrect Mask"
|
14 |
+
}
|
15 |
+
|
16 |
+
# Load model
|
17 |
+
def get_model(num_classes=4):
|
18 |
+
model = fasterrcnn_resnet50_fpn(weights=None)
|
19 |
+
in_features = model.roi_heads.box_predictor.cls_score.in_features
|
20 |
+
model.roi_heads.box_predictor = FastRCNNPredictor(in_features, num_classes)
|
21 |
+
return model
|
22 |
+
|
23 |
+
# Setup
|
24 |
+
device = torch.device("cpu")
|
25 |
+
model = get_model()
|
26 |
+
model.load_state_dict(torch.load("fasterrcnn_resnet50_epoch_4.pth", map_location=device))
|
27 |
+
model.to(device)
|
28 |
+
model.eval()
|
29 |
+
|
30 |
+
# Inference function
|
31 |
+
def predict(image):
|
32 |
+
image_tensor = F.to_tensor(image).unsqueeze(0).to(device)
|
33 |
+
|
34 |
+
with torch.no_grad():
|
35 |
+
prediction = model(image_tensor)
|
36 |
+
|
37 |
+
boxes = prediction[0]["boxes"]
|
38 |
+
labels = prediction[0]["labels"]
|
39 |
+
scores = prediction[0]["scores"]
|
40 |
+
|
41 |
+
draw = ImageDraw.Draw(image)
|
42 |
+
threshold = 0.5
|
43 |
+
|
44 |
+
for box, label, score in zip(boxes, labels, scores):
|
45 |
+
if score > threshold:
|
46 |
+
x1, y1, x2, y2 = box
|
47 |
+
class_name = COCO_CLASSES.get(label.item(), "Unknown")
|
48 |
+
draw.rectangle([x1, y1, x2, y2], outline="red", width=3)
|
49 |
+
draw.text((x1, y1), f"{class_name} ({score:.2f})", fill="red")
|
50 |
+
|
51 |
+
return image
|
52 |
+
|
53 |
+
# Gradio Interface
|
54 |
+
gr.Interface(
|
55 |
+
fn=predict,
|
56 |
+
inputs=gr.Image(type="pil", label="Upload a Face Image"),
|
57 |
+
outputs=gr.Image(type="pil", label="Detection Result"),
|
58 |
+
title="Face Mask Detection - Faster R-CNN",
|
59 |
+
description="Detects faces with mask, without mask, or incorrectly worn mask."
|
60 |
+
).launch()
|