Update app.py
Browse files
app.py
CHANGED
@@ -1,4 +1,38 @@
|
|
1 |
import gradio as gr
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
|
3 |
def greet(name):
|
4 |
return "Hello " + name + "!!"
|
@@ -10,8 +44,8 @@ DETR model finetuned on "anindya64/hardhat" for hard hats detection.
|
|
10 |
|
11 |
demo = gr.Interface(
|
12 |
fn=greet,
|
13 |
-
inputs="
|
14 |
-
outputs="
|
15 |
title=title,
|
16 |
description=description
|
17 |
)
|
|
|
1 |
import gradio as gr
|
2 |
+
from PIL import Image, ImageDraw
|
3 |
+
from transformers import pipeline
|
4 |
+
|
5 |
+
|
6 |
+
def plot_results(image, results, threshold=0.7):
|
7 |
+
image = Image.fromarray(np.uint8(image))
|
8 |
+
draw = ImageDraw.Draw(image)
|
9 |
+
for result in results:
|
10 |
+
score = result["score"]
|
11 |
+
label = result["label"]
|
12 |
+
box = list(result["box"].values())
|
13 |
+
if score > threshold:
|
14 |
+
x, y, x2, y2 = tuple(box)
|
15 |
+
draw.rectangle((x, y, x2, y2), outline="red", width=1)
|
16 |
+
draw.text((x, y), label, fill="white")
|
17 |
+
draw.text(
|
18 |
+
(x + 0.5, y - 0.5),
|
19 |
+
text=str(score),
|
20 |
+
fill="green" if score > 0.7 else "red",
|
21 |
+
)
|
22 |
+
return image
|
23 |
+
|
24 |
+
def predict(image):
|
25 |
+
# make the object detection pipeline
|
26 |
+
obj_detector = pipeline(
|
27 |
+
"object-detection", model="anindya64/detr-resnet-50-dc5-hardhat-finetuned"
|
28 |
+
)
|
29 |
+
results = obj_detector(train_dataset[0]["image"])
|
30 |
+
return plot_results(image)
|
31 |
+
|
32 |
+
|
33 |
+
|
34 |
+
results = obj_detector(image)
|
35 |
+
plot_results(image, results)
|
36 |
|
37 |
def greet(name):
|
38 |
return "Hello " + name + "!!"
|
|
|
44 |
|
45 |
demo = gr.Interface(
|
46 |
fn=greet,
|
47 |
+
inputs=gr.Image(type="filepath", label="Input Image"),
|
48 |
+
outputs="image",
|
49 |
title=title,
|
50 |
description=description
|
51 |
)
|