Repository navigation
Expand file tree
/
Copy pathbeginner.html
More file actions
358 lines (338 loc) · 20.2 KB
/
Copy pathbeginner.html
File metadata and controls
358 lines (338 loc) · 20.2 KB
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
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
<!DOCTYPE html>
<html>
<head>
<title>Jian ZHAO (赵健)</title>
<!-- Custom Theme files -->
<link rel="icon" href="images/favicon.png">
<meta name="viewport" content="width=device-width, initial-scale=1">
<meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
<meta name="keywords" content=""/>
<script type="application/x-javascript"> addEventListener("load", function () {
setTimeout(hideURLbar, 0);
}, false);
function hideURLbar() {
window.scrollTo(0, 1);
} </script>
<!-- //Custom Theme files -->
<link href="css/bootstrap.css" type="text/css" rel="stylesheet" media="all">
<link href="css/style.css" type="text/css" rel="stylesheet" media="all">
<link rel="stylesheet" href="css/lightbox.css">
<!-- js -->
<script src="js/jquery-1.11.1.min.js"></script>
<!-- //js -->
<!--web-fonts-->
<link href='https://fonts.googleapis.com/css?family=Roboto+Condensed:400,300,300italic,400italic,700,700italic'
rel='stylesheet' type='text/css'>
<link href='https://fonts.googleapis.com/css?family=Dosis:400,200,300,500,600,700,800' rel='stylesheet'
type='text/css'>
<!--//web-fonts-->
<!--circle-chart-->
<script src="js/jquery.circlechart.js"></script>
<!--circle-chart-->
</head>
<body>
<div class="body-content">
<!--header-->
<div class="header">
<nav class="navbar navbar-default">
<div class="container">
<div class="navbar-header navbar-left">
<h1><a href="index.html"><img style="width: 40rem" src="images/EVOL_new.jpg"/></a></h1>
</div>
<!--navigation-->
<button type="button" class="navbar-toggle collapsed" data-toggle="collapse"
data-target="#bs-example-navbar-collapse-1">
<span class="sr-only">Toggle navigation</span>
<span class="icon-bar"></span>
<span class="icon-bar"></span>
<span class="icon-bar"></span>
</button>
<div class="header-right">
<div class="top-nav-text">
</div>
<!-- Collect the nav links, forms, and other content for toggling -->
<div class="collapse navbar-collapse navbar-right" id="bs-example-navbar-collapse-1">
<ul class="nav navbar-nav navbar-left">
<li><a href="index.html" class="scroll">Home</a></li>
<li><a href="publication.html" class="scroll">Publications</a></li>
<li><a href="people.html" class="scroll">Members</a></li>
<li><a href="seminar.html" class="scroll">Seminar</a></li>
<li><a href="opening.html" class="scroll">Join Us</a></li>
</ul>
<div class="clearfix"></div>
</div><!--//navigation-->
</div>
<div class="clearfix"></div>
</div>
</nav>
</div>
<!--//header-->
<!--team-->
<div class="container">
<div class="row">
<div class="col-md-12" style="margin-bottom: 20px">
<!-- <h4 class="title">Beginner's Guide to Learn Computer Vision</h4> -->
<h4 class="title">Book and Course</h4>
<div class="contents">
<p>
There is a quick start guide for junior students to master basic Deep Learning knowledge.
We prefer belowing teaching materials for beginner.
</p>
<ol style="margin-bottom: 0px;">
<li>
Dive into Deep Learning.
<a href="https://d2l.ai/">English Edition,</a>
<a href="https://zh.d2l.ai/">Chinese Edition,</a>
<a href="https://space.bilibili.com/1567748478/channel/seriesdetail?sid=358497">Teaching Video Channel.</a>
</li>
<li>
<a href="https://www.deeplearningbook.org/">Deep Learning.</a>
</li>
<li>
<a id="DL_w_pytorch" href="https://isip.piconepress.com/courses/temple/ece_4822/resources/books/Deep-Learning-with-PyTorch.pdf">Deep Learning with Pytorch.</a>
</li>
<li>
<a href="http://cs231n.stanford.edu/">Stanford CS231n: Deep Learning for Computer Vision.</a>
</li>
</ol>
<p>
Besides the teaching materials mentioned above, the docs of popular deep learning frameworks are also excellent supplementary materials.
</p>
<br>
<div class="beginner-img-block">
<img src="images/d2l.png" />
<img src="images/DeepLearning.jpeg">
<img src="images/dlp.jpeg" />
<img src="images/CS231N.png" />
<!-- <img src="images/d2l.png" height="400" width="300" alt="" />
<img src="images/dlp.jpeg" height="400" width="300" alt="" />
<img src="images/CS231N.png" height="400" width="300" alt="" /> -->
</div>
</div>
<h3 class="normal-margin">How to read <i>Dive into Deep Learning</i>?</h3>
<div class="contents">
<p>
This is the
<a href="./images/Dive_into_DL_chapter.PNG" target="_blank">Book Structure</a> of the latest version of Dive into Deep Learning.
For the research of computer vision,
we suggest that beginners can only focus on certain chapters, i.e., 1, 2, 3, 4, 5, 6, 7, 13, 10.
The beginners should try their best to implement the Pytorch code given in this book.
</p>
</div>
<h3 class="normal-margin">How to learn Stanford CS231n?</h3>
<div class="contents">
<p>
CS231N is a famous deep learning course held by Feifei Li,
the sponsor and author of the first large-scale visual dataset ImageNet.
In this course, you can get familiar with multiple basic areas of computer vision,
including image classification, localization, and detection.
This course is a deep dive into details of deep learning architectures with a focus on learning end-to-end models for these tasks,
particularly image classification.
You will learn to implement and train your own neural networks and gain a detailed understanding of cutting-edge research in computer vision.
</p>
<p>
You can get the course videos at Bilibili or Youtube.
Also, you can get detailed course materials on the CS231N official website, mainly supplementary course materials.
If you have enough spare time, accomplishing the assessments is a meaningful and valuable trial to improve your ability to implement neural networks.
</p>
</div>
<h4 class="title">Coding</h4>
<h3 class="normal-margin">Start deep learning with Pytorch.</h3>
<div class="contents">
<p>
<a href="#DL_w_pytorch"><i>Deep Learning with Pytorch</i></a> is a helpful reference book for all beginners who want to be proficient in
using PyTorch to implement their own neural network. But, remember it is only a reference manual,
don't spend too much time reading it but writing code as it tells you. Take it as a technical doc for the PyTorch library,
you can also check the usage on the <a href="https://pytorch.org/docs/stable/index.html">official website</a>.
</p>
</div>
<h3 class="normal-margin">Be familiar with Detectron2.</h3>
<div class="contents">
<p>
<a href="https://github.com/facebookresearch/detectron2">Detectron2</a> includes high-quality implementations of state-of-the-art object detection,
semantic segmentation, and instance segmentation algorithms, implemented based on the PyTorch library.
If you are interested in the above topics, try to use detectron2 to implement your detection or segmentation algorithms.
Detectron2 may be quite difficult for juniors, due to many complex programming mechanisms. When you are familiar with Detectron2,
you will find that implementing a complex vision algorithm is no longer a challenge.
</p>
</div>
<h3 class="normal-margin">How to apply free GPUs for usage?</h3>
<div class="contents">
<p>
Now, it's very convenient to use a free GPU computing platform to run setup deep learning experiments for beginners.
Available GPU computing power providers include Google Colab, Amazon SageMaker, Baidu PaddlePaddle AI Studio, and so on.
But you still need to notice the deep learning framework supported by these platforms when using these computing platforms.
In general, if you are not sure which one is best, Google Colab is a good choice for quick starting.
</p>
<ol style="margin-bottom: 0px;">
<li>
<a href="https://colab.research.google.com/?hl=zh-cn">Google Colab has released a official guidance for users.</a>
</li>
<li>
<a href="https://bfo0rpfl4y.feishu.cn/docs/doccn1B7HSxw6f0pAZo63K80DLd">Instruction of Amazon SageMaker Studio Lab</a>
</li>
</ol>
</div>
<h3 class="normal-margin">Manage your code with Git</h3>
<div class="contents">
<p>
Git is a free and open source distributed version control system designed to handle everything from small to very large projects with speed and efficiency.
Git is <a href="https://git-scm.com/doc">easy to learn</a> and has a tiny footprint with lightning fast performance.
It outclasses SCM tools like Subversion, CVS, Perforce, and ClearCase with features like cheap local branching,
convenient staging areas, and multiple workflows.
</p>
<p>
Besides the official website above, you can also learn git operation from <a href="http://pcottle.github.io/learnGitBranching/">LearnGitBranching</a>.
It is a git repository visualizer, sandbox, and a series of educational tutorials and challenges.
Its primary purpose is to help developers understand git through the power of visualization
(something that's absent when working on the command line).
This is achieved through a game with different levels to get acquainted with the different git commands.
</p>
</div>
<h4 class="title">Articles</h4>
<h3 class="normal-margin">Classic Papers</h3>
<div class="contents">
<p>
Reading the above deep learning materials can help you form understanding of deep learning basics, besides, we recommend following papers to help you further build the transition from deep learning to modern computer vision techniques.
The following papers cover several topics of computer vision, and all are milestones of their research area. Through these papers, you can establish a basic understanding of main vision tasks and quickly carry out research work in related fields.
</p>
<ol>
<li>
Basic Network
<ul>
<li>[AlexNet] Imagenet classification with deep convolutional neural networks <a href="https://proceedings.neurips.cc/paper/2012/file/c399862d3b9d6b76c8436e924a68c45b-Paper.pdf">[NeurIPS link]</a></li>
<li>[ResNet] Deep Residual Learning for Image Recognition <a href="https://arxiv.org/abs/1512.03385">[Arxiv]</a></li>
<li>[Transformer] Attention Is All You Need <a href="http://arxiv.org/abs/1706.03762">[Arxiv]</a></li>
<li>[ViT] An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale <a href="http://arxiv.org/abs/2010.11929">[Arxiv]</a></li>
<li>[Swin] Swin Transformer: Hierarchical Vision Transformer using Shifted Windows <a href="http://arxiv.org/abs/2103.14030">[Arxiv]</a></li>
</ul>
</li>
<li>
Object Detection
<ul>
<li>[Fast R-CNN] Fast R-CNN <a href="https://arxiv.org/abs/1504.08083">[Arxiv]</a></li>
<li>[Faster R-CNN] Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks<a href="https://arxiv.org/abs/1506.01497">[Arxiv]</a></li>
<li>[YOLO] You Only Look Once: Unified, Real-Time Object Detection <a href="https://arxiv.org/abs/1506.02640">[Arxiv]</a></li>
<li>[FCOS] FCOS: Fully Convolutional One-Stage Object Detection <a href="https://arxiv.org/abs/1904.01355">[Arxiv]</a></li>
<li>[DETR] End-to-End Object Detection with Transformers <a href="https://arxiv.org/abs/2005.12872">[Arxiv]</a></li>
</ul>
</li>
<li>
Image Segmentation
<ul>
<li>[FCN] Fully Convolutional Networks for Semantic Segmentation <a href="https://arxiv.org/abs/1411.4038">[Arxiv]</a></li>
<li>[U-Net] U-Net: Convolutional Networks for Biomedical Image Segmentation <a href="https://arxiv.org/abs/1505.04597">[Arxiv]</a></li>
<li>[DeepLab] DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs <a href="https://arxiv.org/abs/1606.00915">[Arxiv]</a></li>
<li>[PSPNet] Pyramid Scene Parsing Network <a href="http://arxiv.org/abs/1612.01105">[Arxiv]</a></li>
<li>[Mask R-CNN] Mask R-CNN <a href="https://arxiv.org/abs/1703.06870">[Arxiv]</a></li>
<li>[Non-local] Non-local Neural Networks <a href="http://arxiv.org/abs/1711.07971">[Arxiv]</a></li>
<li>[CCNet] CCNet: Criss-Cross Attention for Semantic Segmentation <a href="http://arxiv.org/abs/1811.11721">[Arxiv]</a></li>
</ul>
</li>
</ol>
</div>
<h3 class="normal-margin">How to acquire cutting-edge papers in computer vision?</h3>
<div class="contents">
<p>
<a href="https://openaccess.thecvf.com/menu">CVF Open Access</a>
is a non-profit organization whose purpose is to foster and support research on all aspects of computer vision,
including through supporting such conferences as Computer Vision and Pattern Recognition (CVPR)
and the International Conference on Computer Vision (ICCV).
The CVF can solicit donations and provide grants in furtherance of its goals.
</p>
<p>
<a href="https://paperswithcode.com/">Papers with Code</a> is the platform that contains research papers
with code implementations by the authors or community.
Recently, Papers with Code have grown in both popularity and in terms of providing a complete ecosystem for machine learning research.
</p>
</div>
<h3 class="normal-margin">How to be a qualified Ph.D. or super Master?</h3>
<div class="contents">
<p>
Please refer to the <a href="https://weiyc.github.io/seminar/How_to_be_a_qualified_PhD_tianlong.pdf">slides</a> given by Tianlong.
</p>
</div>
<h3 class="normal-margin">Be knowledgeable about top-tier computer vision venues</h3>
<div class="contents">
<p>
Top-tier Conferences
<ol style="list-style: circle">
<li>IEEE Computer Vision and Pattern Recognition (CVPR)</li>
<li>IEEE International Conference on Computer Vision (ICCV)</li>
<li>European Conference on Computer Vision (ECCV)</li>
<li>Neural Information Processing Systems (NeurIPS)</li>
<li>International Conference on Learning Representations (ICLR)</li>
<li>International Conference on Machine Learning (ICML)</li>
</ol>
</p>
<p>
Top-tier Journals
</p>
<ol style="list-style: circle">
<li>IEEE Transactions on Pattern Analysis and Machine Intelligence(TPAMI)</li>
<li>International Journal of Computer Vision (IJCV)</li>
<li>IEEE Transactions on Image Processing (TIP)</li>
</ol>
<p>
In WEI Lab, we only concentrate on publishing papers in the above-listed venues.
</p>
</div>
<h3 class="normal-margin">Resources in Zhihu</h3>
<div class="contents">
<p>
Zhihu is a great comprehensive forum in China, you can find a variety of interesting opinions about various events.
Besides, it's worth noticing that there are also plentiful deep learning and computer vision articles,
including beginning tutorials, technology sharing, and cutting-edge paper sharing.
The most exciting thing is you can find some Chinese scholars sharing their thoughts about their professional fields.
</p>
</div>
</div>
</div>
</div>
<!--footer-->
<div class="footer">
<div class="container">
<div class="col-md-5">
<h1><img style="width: 15rem" src="images/TeleAI.jpg"/></a></h1>
<p style="margin-top:25px;;text-align:center">Modified: 18 April 2024</p>
<!-- <p>Copyright © 2021.Company name All rights reserved. </p> -->
</div>
<div class="col-md-5">
<div style="display:inline-block;width:300px;"><script type='text/javascript' id='mapmyvisitors' src='https://mapmyvisitors.com/map.js?cl=ffffff&w=a&t=tt&d=hcO8hctdFbPEhVELE4et7rGqxoXEOAOgkmdpwJyBOoQ'></script></div>
</div>
</div>
<!--//footer-->
<!--banner Slider starts Here-->
<script src="js/responsiveslides.min.js"></script>
<script>
// You can also use "$(window).load(function() {"
$(function () {
// Slideshow 3
$("#slider3").responsiveSlides({
auto: true,
pager: true,
nav: false,
speed: 500,
namespace: "callbacks",
before: function () {
$('.events').append("<li>before event fired.</li>");
},
after: function () {
$('.events').append("<li>after event fired.</li>");
}
});
});
</script>
<!--//End-slider-script-->
<!-- start-smooth-scrolling-->
<script type="text/javascript" src="js/move-top.js"></script>
<script type="text/javascript" src="js/easing.js"></script>
<!--//smooth-scrolling-of-move-up-->
<!-- Bootstrap core JavaScript
================================================== -->
<!-- Placed at the end of the document so the pages load faster -->
<script src="js/bootstrap.js"></script>
</div>
</body>
</body>
</html>