nickmuchi commited on
Commit
6f3b044
1 Parent(s): 8153ff2

Create app.py

Browse files
Files changed (1) hide show
  1. app.py +153 -0
app.py ADDED
@@ -0,0 +1,153 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import io
2
+ import gradio as gr
3
+ import matplotlib.pyplot as plt
4
+ import requests, validators
5
+ import torch
6
+ import pathlib
7
+ from PIL import Image
8
+ from transformers import AutoFeatureExtractor, DetrForObjectDetection, YolosForObjectDetection
9
+
10
+ import os
11
+
12
+ # colors for visualization
13
+ COLORS = [
14
+ [0.000, 0.447, 0.741],
15
+ [0.850, 0.325, 0.098],
16
+ [0.929, 0.694, 0.125],
17
+ [0.494, 0.184, 0.556],
18
+ [0.466, 0.674, 0.188],
19
+ [0.301, 0.745, 0.933]
20
+ ]
21
+
22
+ def make_prediction(img, feature_extractor, model):
23
+ inputs = feature_extractor(img, return_tensors="pt")
24
+ outputs = model(**inputs)
25
+ img_size = torch.tensor([tuple(reversed(img.size))])
26
+ processed_outputs = feature_extractor.post_process(outputs, img_size)
27
+ return processed_outputs[0]
28
+
29
+ def fig2img(fig):
30
+ buf = io.BytesIO()
31
+ fig.savefig(buf)
32
+ buf.seek(0)
33
+ img = Image.open(buf)
34
+ return img
35
+
36
+
37
+ def visualize_prediction(pil_img, output_dict, threshold=0.7, id2label=None):
38
+ keep = output_dict["scores"] > threshold
39
+ boxes = output_dict["boxes"][keep].tolist()
40
+ scores = output_dict["scores"][keep].tolist()
41
+ labels = output_dict["labels"][keep].tolist()
42
+ if id2label is not None:
43
+ labels = [id2label[x] for x in labels]
44
+
45
+ plt.figure(figsize=(16, 10))
46
+ plt.imshow(pil_img)
47
+ ax = plt.gca()
48
+ colors = COLORS * 100
49
+ for score, (xmin, ymin, xmax, ymax), label, color in zip(scores, boxes, labels, colors):
50
+ ax.add_patch(plt.Rectangle((xmin, ymin), xmax - xmin, ymax - ymin, fill=False, color=color, linewidth=3))
51
+ ax.text(xmin, ymin, f"{label}: {score:0.2f}", fontsize=15, bbox=dict(facecolor="yellow", alpha=0.5))
52
+ plt.axis("off")
53
+ return fig2img(plt.gcf())
54
+
55
+ def detect_objects(model_name,url_input,image_input,threshold):
56
+
57
+ #Extract model and feature extractor
58
+ feature_extractor = AutoFeatureExtractor.from_pretrained(model_name)
59
+
60
+ if 'detr' in model_name:
61
+
62
+ model = DetrForObjectDetection.from_pretrained(model_name)
63
+
64
+ elif 'yolos' in model_name:
65
+
66
+ model = YolosForObjectDetection.from_pretrained(model_name)
67
+
68
+ if validators.url(url_input):
69
+ image = Image.open(requests.get(url_input, stream=True).raw)
70
+
71
+ elif image_input:
72
+ image = image_input
73
+
74
+ #Make prediction
75
+ processed_outputs = make_prediction(image, feature_extractor, model)
76
+
77
+ #Visualize prediction
78
+ viz_img = visualize_prediction(image, processed_outputs, threshold, model.config.id2label)
79
+
80
+ return viz_img
81
+
82
+ def set_example_image(example: list) -> dict:
83
+ return gr.Image.update(value=example[0])
84
+
85
+ def set_example_url(example: list) -> dict:
86
+ return gr.Textbox.update(value=example[0])
87
+
88
+
89
+ title = """<h1 id="title">Object Detection App with DETR and YOLOS</h1>"""
90
+
91
+ description = """
92
+ Links to HuggingFace Models:
93
+ - [facebook/detr-resnet-50](https://huggingface.co/facebook/detr-resnet-50)
94
+ - [facebook/detr-resnet-101](https://huggingface.co/facebook/detr-resnet-101)
95
+ - [hustvl/yolos-small](https://huggingface.co/hustvl/yolos-small)
96
+ - [hustvl/yolos-tiny](https://huggingface.co/hustvl/yolos-tiny)
97
+ """
98
+
99
+ models = ["facebook/detr-resnet-50","facebook/detr-resnet-101",'hustvl/yolos-small','hustvl/yolos-tiny']
100
+ urls = ["https://c8.alamy.com/comp/J2AB4K/the-new-york-stock-exchange-on-the-wall-street-in-new-york-J2AB4K.jpg"]
101
+
102
+ twitter_link = """
103
+ [![](https://img.shields.io/twitter/follow/nickmuchi?label=@nickmuchi&style=social)](https://twitter.com/nickmuchi)
104
+ """
105
+
106
+ css = '''
107
+ h1#title {
108
+ text-align: center;
109
+ }
110
+ '''
111
+ demo = gr.Blocks(css=css)
112
+
113
+ with demo:
114
+ gr.Markdown(title)
115
+ gr.Markdown(description)
116
+ gr.Markdown(twitter_link)
117
+ options = gr.Dropdown(choices=models,label='Select Object Detection Model',show_label=True)
118
+ slider_input = gr.Slider(minimum=0.2,maximum=1,value=0.7,label='Prediction Threshold')
119
+
120
+ with gr.Tabs():
121
+ with gr.TabItem('Image URL'):
122
+ with gr.Row():
123
+ url_input = gr.Textbox(lines=2,label='Enter valid image URL here..')
124
+ img_output_from_url = gr.Image(shape=(650,650))
125
+
126
+ with gr.Row():
127
+ example_url = gr.Dataset(components=[url_input],samples=[[str(url)] for url in urls])
128
+
129
+ url_but = gr.Button('Detect')
130
+
131
+ with gr.TabItem('Image Upload'):
132
+ with gr.Row():
133
+ img_input = gr.Image(type='pil')
134
+ img_output_from_upload= gr.Image(shape=(650,650))
135
+
136
+ with gr.Row():
137
+ example_images = gr.Dataset(components=[img_input],
138
+ samples=[[path.as_posix()]
139
+ for path in sorted(pathlib.Path('images').rglob('*.JPG'))])
140
+
141
+ img_but = gr.Button('Detect')
142
+
143
+
144
+ url_but.click(detect_objects,inputs=[options,url_input,img_input,slider_input],outputs=img_output_from_url,queue=True)
145
+ img_but.click(detect_objects,inputs=[options,url_input,img_input,slider_input],outputs=img_output_from_upload,queue=True)
146
+ example_images.click(fn=set_example_image,inputs=[example_images],outputs=[img_input])
147
+ example_url.click(fn=set_example_url,inputs=[example_url],outputs=[url_input])
148
+
149
+
150
+ gr.Markdown("![visitor badge](https://visitor-badge.glitch.me/badge?page_id=nickmuchi-object-detection-with-detr-and-yolos)")
151
+
152
+
153
+ demo.launch(enable_queue=True)