Spaces:
Runtime error
Runtime error
Commit
•
c72f916
1
Parent(s):
099822b
Upload app.py
Browse files
app.py
ADDED
@@ -0,0 +1,117 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
import torch
|
3 |
+
|
4 |
+
from transformers import AutoImageProcessor, AutoModelForObjectDetection
|
5 |
+
#from transformers import pipeline
|
6 |
+
|
7 |
+
from PIL import Image
|
8 |
+
import matplotlib.pyplot as plt
|
9 |
+
import matplotlib.patches as patches
|
10 |
+
|
11 |
+
import io
|
12 |
+
from random import choice
|
13 |
+
|
14 |
+
|
15 |
+
image_processor_tiny = AutoImageProcessor.from_pretrained("/home/pc-1/Documents/Object detection/weights")
|
16 |
+
model_tiny = AutoModelForObjectDetection.from_pretrained("/home/pc-1/Documents/Object detection/weights")
|
17 |
+
|
18 |
+
import gradio as gr
|
19 |
+
|
20 |
+
|
21 |
+
COLORS = ["#ff7f7f", "#ff7fbf", "#ff7fff", "#bf7fff",
|
22 |
+
"#7f7fff", "#7fbfff", "#7fffff", "#7fffbf",
|
23 |
+
"#7fff7f", "#bfff7f", "#ffff7f", "#ffbf7f"]
|
24 |
+
|
25 |
+
fdic = {
|
26 |
+
"family" : "DejaVu Serif",
|
27 |
+
"style" : "normal",
|
28 |
+
"size" : 18,
|
29 |
+
"color" : "yellow",
|
30 |
+
"weight" : "bold"
|
31 |
+
}
|
32 |
+
|
33 |
+
|
34 |
+
def get_figure(in_pil_img, in_results):
|
35 |
+
plt.figure(figsize=(16, 10))
|
36 |
+
plt.imshow(in_pil_img)
|
37 |
+
ax = plt.gca()
|
38 |
+
|
39 |
+
for score, label, box in zip(in_results["scores"], in_results["labels"], in_results["boxes"]):
|
40 |
+
selected_color = choice(COLORS)
|
41 |
+
|
42 |
+
box_int = [i.item() for i in torch.round(box).to(torch.int32)]
|
43 |
+
x, y, w, h = box_int[0], box_int[1], box_int[2]-box_int[0], box_int[3]-box_int[1]
|
44 |
+
#x, y, w, h = torch.round(box[0]).item(), torch.round(box[1]).item(), torch.round(box[2]-box[0]).item(), torch.round(box[3]-box[1]).item()
|
45 |
+
|
46 |
+
ax.add_patch(plt.Rectangle((x, y), w, h, fill=False, color=selected_color, linewidth=3, alpha=0.8))
|
47 |
+
ax.text(x, y, f"{model_tiny.config.id2label[label.item()]}: {round(score.item()*100, 2)}%", fontdict=fdic, alpha=0.8)
|
48 |
+
|
49 |
+
plt.axis("off")
|
50 |
+
|
51 |
+
return plt.gcf()
|
52 |
+
|
53 |
+
|
54 |
+
def infer(in_pil_img, in_model="yolos-tiny", in_threshold=0.9):
|
55 |
+
target_sizes = torch.tensor([in_pil_img.size[::-1]])
|
56 |
+
inputs = image_processor_tiny(images=in_pil_img, return_tensors="pt")
|
57 |
+
outputs = model_tiny(**inputs)
|
58 |
+
|
59 |
+
# convert outputs (bounding boxes and class logits) to COCO API
|
60 |
+
results = image_processor_tiny.post_process_object_detection(outputs, threshold=in_threshold, target_sizes=target_sizes)[0]
|
61 |
+
|
62 |
+
figure = get_figure(in_pil_img, results)
|
63 |
+
|
64 |
+
buf = io.BytesIO()
|
65 |
+
figure.savefig(buf, bbox_inches='tight')
|
66 |
+
buf.seek(0)
|
67 |
+
output_pil_img = Image.open(buf)
|
68 |
+
|
69 |
+
return output_pil_img
|
70 |
+
|
71 |
+
|
72 |
+
with gr.Blocks(title="YOLOS Object Detection - ClassCat",
|
73 |
+
css=".gradio-container {background:lightyellow;}"
|
74 |
+
) as demo:
|
75 |
+
#sample_index = gr.State([])
|
76 |
+
|
77 |
+
gr.HTML("""<div style="font-family:'Times New Roman', 'Serif'; font-size:16pt; font-weight:bold; text-align:center; color:royalblue;">YOLOS Object Detection</div>""")
|
78 |
+
|
79 |
+
gr.HTML("""<h4 style="color:navy;">1. Select a model.</h4>""")
|
80 |
+
|
81 |
+
model = gr.Radio(["yolos-tiny", "yolos-small"], value="yolos-tiny", label="Model name")
|
82 |
+
|
83 |
+
gr.HTML("""<br/>""")
|
84 |
+
gr.HTML("""<h4 style="color:navy;">2-a. Select an example by clicking a thumbnail below.</h4>""")
|
85 |
+
gr.HTML("""<h4 style="color:navy;">2-b. Or upload an image by clicking on the canvas.</h4>""")
|
86 |
+
|
87 |
+
with gr.Row():
|
88 |
+
input_image = gr.Image(label="Input image", type="pil")
|
89 |
+
output_image = gr.Image(label="Output image with predicted instances", type="pil")
|
90 |
+
|
91 |
+
gr.Examples(['samples/cats.jpg', 'samples/detectron2.png', 'samples/cat.jpg', 'samples/hotdog.jpg'], inputs=input_image)
|
92 |
+
|
93 |
+
gr.HTML("""<br/>""")
|
94 |
+
gr.HTML("""<h4 style="color:navy;">3. Set a threshold value (default to 0.9)</h4>""")
|
95 |
+
|
96 |
+
threshold = gr.Slider(0, 1.0, value=0.9, label='threshold')
|
97 |
+
|
98 |
+
gr.HTML("""<br/>""")
|
99 |
+
gr.HTML("""<h4 style="color:navy;">4. Then, click "Infer" button to predict object instances. It will take about 10 seconds (yolos-tiny) or 20 seconds (yolos-small).</h4>""")
|
100 |
+
|
101 |
+
send_btn = gr.Button("Infer")
|
102 |
+
send_btn.click(fn=infer, inputs=[input_image, model, threshold], outputs=[output_image])
|
103 |
+
|
104 |
+
gr.HTML("""<br/>""")
|
105 |
+
gr.HTML("""<h4 style="color:navy;">Reference</h4>""")
|
106 |
+
gr.HTML("""<ul>""")
|
107 |
+
gr.HTML("""<li><a href="https://huggingface.co/docs/transformers/model_doc/yolos" target="_blank">Hugging Face Transformers - YOLOS</a>""")
|
108 |
+
gr.HTML("""</ul>""")
|
109 |
+
|
110 |
+
|
111 |
+
#demo.queue()
|
112 |
+
demo.launch(debug=True)
|
113 |
+
|
114 |
+
|
115 |
+
|
116 |
+
|
117 |
+
### EOF ###
|