-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathindex.html
218 lines (183 loc) · 11.4 KB
/
index.html
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
207
208
209
210
211
212
213
214
215
216
217
218
<!DOCTYPE html>
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">
<head>
<title>Ao Meng's Homepage</title>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<meta name="author" content="Ao Meng" />
<link rel="stylesheet" href="w3.css">
<style>
.w3-sidebar a {font-family: "Roboto", sans-serif}
body,h1,h2,h3,h4,h5,h6,.w3-wide {font-family: "Montserrat", sans-serif;}
</style>
<link rel="icon" type="image/png" href="images/icons.png">
<!--
<script src="jquery.min.js"></script>
<script>
$(document).ready(function(){
// Add smooth scrolling to all links
$("a").on('click', function(event) {
// Make sure this.hash has a value before overriding default behavior
if (this.hash !== "") {
// Prevent default anchor click behavior
event.preventDefault();
// Store hash
var hash = this.hash;
// Using jQuery's animate() method to add smooth page scroll
// The optional number (800) specifies the number of milliseconds it takes to scroll to the specified area
$('html, body').animate({
scrollTop: $(hash).offset().top
}, 800, function(){
// Add hash (#) to URL when done scrolling (default click behavior)
window.location.hash = hash;
});
} // End if
});
});
</script>
//-->
</head>
<body class="w3-content" style="max-width:1000px">
<!-- Sidebar/menu -->
<nav class="w3-sidebar w3-bar-block w3-black w3-collapse w3-top w3-right" style="z-index:3;width:150px" id="mySidebar">
<div class="w3-container w3-display-container w3-padding-16">
<h3><b>Ao</b></h3>
</div>
<div class="w3-padding-64 w3-text-light-grey w3-large" style="font-weight:bold">
<a href="#home" class="w3-bar-item w3-button">Home</a>
<a href="#news" class="w3-bar-item w3-button">News</a>
<a href="#projects" class="w3-bar-item w3-button">Projects</a>
<a href="#talks" class="w3-bar-item w3-button">Talks</a>
<a href="#publications" class="w3-bar-item w3-button">Research</a>
<a href="#service" class="w3-bar-item w3-button">Services</a>
<a href="#award" class="w3-bar-item w3-button">Awards</a>
<a href="#live" class="w3-bar-item w3-button">Lives</a>
</div>
</nav>
<!-- Top menu on small screens -->
<header class="w3-bar w3-top w3-hide-large w3-black w3-xlarge">
<div class="w3-bar-item w3-padding-24">AO</div>
<a href="javascript:void(0)" class="w3-bar-item w3-button w3-padding-24 w3-right" style="font-stretch: extra-expanded;" onclick="w3_open()"><b>≡</b></a>
</div>
</header>
<!-- Overlay effect when opening sidebar on small screens -->
<div class="w3-overlay w3-hide-large" onclick="w3_close()" style="cursor:pointer" title="close side menu" id="myOverlay"></div>
<!-- !PAGE CONTENT! -->
<div class="w3-main" style="margin-left:150px">
<!-- Push down content on small screens -->
<div class="w3-hide-large" style="margin-top:83px"></div>
<!-- The Home Section -->
<div class="w3-container w3-center w3-padding-32" id="home">
<img style="width: 80%;max-width: 320px" alt="profile photo" src="images/Meng.png">
<h1>Ao Meng</h1>
<p class="w3-justify" style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;max-width:600px">
I am now the postgraduate in automation at <a href="https://www.hitsz.edu.cn/index.html">HITSZ</a>.I am studying in CLASS LAB,
where I am advised by Prof. <a href="https://scholar.google.com.sg/citations?hl=zh-CN&user=hUhUGfYAAAAJ">Liang Hu</a></a>
and working on reinforcement learning, path planning and robot control .
Before that, I did my bachelors at school of HITSZ, also.
</p>
<p class="w3-center">
<a href="[email protected]">Email</a>  / 
<!-- <a href="https://scholar.google.com/citations?user=isizOkYAAAAJ">Google Scholar</a>  /  -->
<!-- <a href="https://www.zhihu.com/people/YunheWang"> Zhi Hu </a>  /  -->
<a href="https://github.com/Ice-mao"> Github </a>
</p>
</tbody></table>
</div>
<!-- The News Section -->
<div class="w3-container w3-light-grey w3-padding-32" id="news">
<h2>News</h2>
<p><li> <a style="color:red">09/2023, update the homepage</a>.</li></p>
<p><li> 09/2023, I join the CLASS LAB formally.</li></p>
<!--
<p><li> 09/2021, 10 papers have been accepted by <a href="https://nips.cc/Conferences/2021">NeurIPS 2021</a>. Congrats to everyone!</li></p>
<p><li> 09/2021, The journal version of versatile filters has been accepted by <a href="https://ieeexplore.ieee.org/iel7/34/4359286/09543586.pdf">IEEE TPAMI</a>.</li></p>
<p><li> 07/2021, 1 paper has been accepted by <a href="https://iccv2021.thecvf.com/home">ICCV 2021</a>.</li></p>
<p><li> 05/2021, 1 paper has been accepted by <a href="https://icml.cc/">ICML 2021</a>.</li></p>
<p><li> 05/2021, I have been selected as a Senior Area Chair for <a href="http://valser.org/">VALSE 2021</a>.</li></p>
<p><li> 03/2021, I accepted the invitation to serve as an Area Chair for <a href="https://nips.cc/Conferences/2021/">NeurIPS 2021</a>.</li></p>
<p><li> 03/2021, 9 papers have been accepted by <a href="http://cvpr2021.theRcvf.com/">CVPR 2021</a>.</li></p>
-->
</div>
<!-- The Projects Section -->
<div class="w3-container w3-padding-32" id="projects">
<h2>Recent Projects</h2>
<p class="w3-justify">
I want to update my projects here.So I shoule strive to improve myself !
</p>
<h4><li>Projects done during undergraduate studies</li></h4>
<img style="width:96%;" src="images/VanillaNet.png">
<p class="w3-justify">
<a style="color:rgb(231, 216, 216)">Robomaster is an interesting race!</a>
I make the dart robot which can shoot darts to hit the opponent's buildings.
</p>
<!-- <img style="width:96%;" src="images/VanillaNet.png">
<p class="w3-justify">
<a style="color: #447ec9" href="https://github.com/huawei-noah/VanillaNet">Project Page</a> | <a style="color: #447ec9" href="https://arxiv.org/abs/2305.12972">Paper</a> | <a style="color: #447ec9" href="https://www.zhihu.com/question/531529633/answer/3047230939">Discussion on Zhihu</a>
</p>
<p class="w3-justify">
<span style="color:red">VanillaNet is remarkable!</span> The concept was born from embracing the "less is more" philosophy in computer vision. It's elegantly designed by avoiding intricate depth and operations, such as self-attention, making it remarkably powerful yet concise. The 6-layer VanillaNet surpasses ResNet-34, and the 13-layer variant achieves about 83% Top-1 accuracy, outpacing the performance of networks with hundreds of layers, and revealing exceptional hardware efficiency advantages.
</p>
<h4><li>Adder Neural Networks</li></h4>
<img style="width:96%;" src="images/AdderNet.jpg">
<p class="w3-justify">
<a style="color: #447ec9" href="https://github.com/huawei-noah/AdderNet">Project Page</a> | <a style="color: #447ec9" href="https://arxiv.org/pdf/2101.10015.pdf">Hardware Implementation</a>
</p>
<p class="w3-justify">
I would like to say, <span style="color:red">AdderNet is very cool!</span> The initial idea was came up in about 2017 when climbing with some friends at Beijing. By replacing all convolutional layers (except the first and the last layers), we now can obtain comparable performance on ResNet architectures. In addition, to make the story more complete, we recent release the hardware implementation and some quantization methods. The results are quite encouraging, we can reduce both <strong>the energy consumption and thecircuit areas significantly without affecting the performance</strong>. Now, we are working on more applications to reduce the costs of launching AI algorithms such as low-level vision, detection, and NLP tasks.
</p>
<h4><li>GhostNet on MindSpore: SOTA Lightweight CV Networks</li></h4>
<img style="width:96%;" src="images/GhostNet.png">
<p class="w3-justify">
<a style="color: #447ec9" href="https://live.huawei.com/huaweiconnect/meeting/cn/5872.html">Huawei Connect (HC) 2020</a> | <a style="color: #447ec9" href="https://www.mindspore.cn/resources/hub">MindSpore Hub</a>
</p>
<p class="w3-justify">
The initial verison of GhostNet was accepted by CVPR 2020, which achieved SOTA performance on ImageNet: <span style="color:red">75.7%</span> top1 acc with only <span style="color:red">226M FLOPS</span>. In the current version, we release a series computer vision models (e.g. int8 quantization, detection, and larger networks) on <strong>MindsSpore 1.0</strong> and <strong>Mate 30 Pro (Kirin 990)</strong>.
</p>
<h4><li>AI on Ascend: Real-Time Video Style Transfer</li></h4>
<img style="width:32%;" src="images/atlas200.png">    <img style="width:60%;" src="images/video.gif">
<p class="w3-justify">
<a style="color: #447ec9" href="https://www.huaweicloud.com/intl/en-us/HDC.Cloud.html">Huawei Developer Conference (HDC) 2020</a> | <a style="color: #447ec9" href="https://developer.huaweicloud.com/exhibition/Atlas_neural_style.html">Online Demo</a>
</p>
<p class="w3-justify">
This project aims to develop a video style transfer system on the <strong>Huawei Atlas 200 DK AI developer Kit</strong>. The latency of the original model for processing one image is about <span style="color:red">630ms</span>. After accelerating it using our method, the lantency now is about <span style="color:red">40ms</span>.
</p> -->
</div>
<!-- The Talks Section -->
<div class="w3-container w3-light-grey w3-padding-32" id="talks">
<h2>Talks</h2>
<!-- <p><li> 12/2022, Hardware Efficient Deep Learning at <a href="https://ccf.org.cn/cncc2022/schedule_d_4179">China National Computer Congress (CNCC) 2022</a>. Thanks Prof. <a href="http://www.nlpr.ia.ac.cn/jcheng/">Jian Cheng</a> for the invitation.</li></p> -->
<!-- The Publications Section -->
<div class="w3-container w3-padding-32"" id="publications">
<h2>Research</h2>
<p class="w3-left-align" style="line-height:200%">
I'm interested in exploring <strong> reinforcement learing and path planning</strong> practicing in autonomous underwater vehicle(AUV)
</p>
<h4> Conference Papers:</h4>
<ol>
<!--
<p>
<li><strong>Title</strong>
<br>
Author, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2021 | <a style="color: #447ec9" href="paper link">paper</a> | <a style="color: #447ec9" href="https://github.com/">code</a> | <span style="color:red">Oral Presentation</span>
</p>
-->
</ol>
</p>
</div>
<!-- The Services Section -->
<div class="w3-container w3-light-grey w3-padding-32" id="service">
<h2>Services</h2>
<p><li> Program Committee Members of ICCV 2021, AAAI 2021, ICLR 2021, NeurIPS 2020, ICML 2020, ECCV 2020, CVPR 2020, ICLR 2020, AAAI 2020, ICCV 2019, CVPR 2019, ICLR 2019, AAAI 2019, IJCAI 2018, AAAI 2018, NeurIPS 2018, etc.</p>
</div>
<!-- The Awards Section -->
<div class="w3-container w3-padding-32" id="award">
<h2>Awards</h2>
<!-- <p><li> 2016, National Scholarship for Graduate Students.</p> -->
</div>
<div class="w3-light-grey w3-center w3-padding-24">
</body>
</html>