PageSourceSearch

https://yunfan1202.github.io/

html yunfan1202.github.io collected 2026-10-03 10:13:42 UTC 24,795 bytes, 467 lines download raw bytes

1<!DOCTYPE HTML>
2<html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
3
4  <title>Yunfan Ye's Homepage</title>
5
6  <meta name="author" content="Yunfan Ye">
7  <meta name="viewport" content="width=device-width, initial-scale=1">
8
9  <link rel="stylesheet" type="text/css" href="stylesheet.css">
10	<link rel="icon" href="data:image/svg+xml,<svg xmlns=%22http://www.w3.org/2000/svg%22 viewBox=%220 0 100 100%22><text y=%22.9em%22 font-size=%2290%22>🌐</text></svg>">
11      <style>
12    .projects-box {
13      /* background-color: green; */
14      height: 160px;
15      position: relative;
16      overflow: hidden;
17      transition: all ease-in .1s;
18
19    }
20    .projects-show {
21      padding: 0;
22      margin: 0;
23      position: absolute;
24      bottom: 0;
25      left: 0;
26      text-align: center;
27      height: 30px;
28      line-height: 20px;
29      width: 100%;
30      font-size: 12px;
31      color: #0067c8;
32      cursor: pointer;
33      background: linear-gradient(to bottom, transparent, #fff, #fff);
34    }
35    .projects-show-text {
36      position: relative;
37      top: 10px;
38    }
39  </style>
40</head>
41
42<body>
43  <table style="width:100%;max-width:800px;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;"><tbody>
44    <tr style="padding:0px">
45      <td style="padding:0px">
46        <table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;"><tbody>
47          <tr style="padding:0px">
48            <td style="padding:2.5%;width:63%;vertical-align:middle">
49              <p style="text-align:center">
50                <name>Yunfan Ye</name>
51              </p>
52
53              <p style="text-align:justify">Yunfan Ye (叶云帆) is an Assistant Professor in <a href="http://design.hnu.edu.cn/gy.htm">School of Design</a>, <a href="https://www.hnu.edu.cn/">Hunan University (HNU)</a>, China.
54                I earned my Ph.D. degree in December 2023 in <a href="https://www.nudt.edu.cn/">National University of Defense Technology</a>, under the supervision of <a href="http://individual.utoronto.ca/zcai/">Prof. Zhiping Cai</a> and <a href="https://kevinkaixu.net/">Prof. Kai Xu</a> in <a href="https://kevinkaixu.net/group.html">iGrape Lab</a>.
55                I got my Master's degree in Computer Science in 2019 from  <a href="https://www.stevens.edu/">Stevens Institute of Technology</a>, and Bachelor's degree in Computer Science in 2017 from <a href="https://www.xmu.edu.cn/">Xiamen University</a>, China.
56
57              </p>
58
59              <p style="text-align:center">
60                <a href="mailto:[email protected]">Email</a> &nbsp/&nbsp
61<!--                <a href="data/Yunfan-CV.pdf">CV</a> &nbsp/&nbsp-->
62<!--                <a href="data/Yunfan-bio.txt">Bio</a> &nbsp/&nbsp-->
63                <a href="https://scholar.google.com/citations?user=iTGg6eQAAAAJ">Google Scholar</a> &nbsp/&nbsp
64                <a href="https://github.com/yunfan1202/">Github</a>
65
66              </p>
67            </td>
68            <td style="padding:2.5%;width:40%;max-width:40%">
69              <a href="images/yeyunfan-photo.jpg"><img style="width:65%;max-width:65%" alt="profile photo" src="images/yeyunfan-photo.jpg" class="hoverZoomLink"></a>
70            </td>
71          </tr>
72        </tbody></table>
73
74         <!--
75         <table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;"><tbody>
76            <tr>
77            <td style="padding:20px;width:100%;vertical-align:middle">
78              <heading>News</heading>
79              <ul class="projects-box" id="projects-box">
80	      <li><b>[<font color="red">2024.01</font>]</b> One paper has been accepted by <em>IEEE TIP</em>.</li>
81
82                  <p class="projects-show" id="projects-show"><span class="projects-show-text" id="projects-show-text">More</span></p>
83
84              </ul>
85            </td>
86          </tr>
87        </tbody></table>
88        -->
89        <table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;"><tbody>
90            <tr>
91            <td style="padding:20px;width:100%;vertical-align:middle">
92              <heading>Teaching</heading>
93              <p>
94                智能设计方法 (Intelligent Design Method) <a href="https://github.com/yunfan1202/intellegent_design">[Code]</a>
95              </p>
96              <p>
97                数据可视化 (Data Visualization) <a href="https://yunfan1202.github.io/data_visualization/fall2025/">[Project Page]</a>
98              </p>
99            </td>
100          </tr>
101        </tbody></table>
102
103
104        <table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;"><tbody>
105            <tr>
106            <td style="padding:20px;width:100%;vertical-align:middle">
107              <heading>Research</heading>
108              <p>
109                My research interest mainly include computer vision and graphics, intelligent design and their applications, especially edge detection, neural radiance field. The representative papers are <span class="highlight">highlighted</span>.
110              </p>
111            </td>
112          </tr>
113        </tbody></table>
114
115
116<!---------------------------------------------HumanSAM : Classifying Human-centric Forgery Videos in Human Spatial, Appearance, and Motion Anomaly------------------------------------------------------------------------------>
117        <table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;"><tbody>
118
119<!--              -->
120            <td style="padding:20px;width:25%;vertical-align:middle">
121              <div class="one">
122                <img src='images/HumanSAM.jpg' height= 70%>
123              </div>
124            </td>
125            <td style="padding:20px;width:75%;vertical-align:middle">
126              <a href="https://dejian-lc.github.io/humansam/">
127                <papertitle>HumanSAM : Classifying Human-centric Forgery Videos in Human Spatial, Appearance, and Motion Anomaly</papertitle>
128              </a>
129              <br>
130              Chang Liu*,
131              <strong>Yunfan Ye*</strong>,
132              Fan Zhang,
133              Qingyang Zhou,
134              Yuchuan Luo,
135              <a href="http://individual.utoronto.ca/zcai/">Zhiping Cai</a>
136              <br>
137              <em>ICCV</em>, 2025
138              <br>
139              <a href="https://dejian-lc.github.io/humansam/">[Project Page]</a>
140              <a href="https://arxiv.org/abs/2507.19924">[Paper]</a>
141              <!--
142              <a href="https://github.com/GuHuangAI/DiffusionEdge">[Code]</a>
143              -->
144              <p></p>
145              <p style="text-align:justify">
146                Numerous synthesized videos from generative models, especially human-centric ones that simulate
147                realistic human actions, pose significant threats to human information security and authenticity.
148                While progress has been made in binary forgery video detection, the lack of fine-grained understanding
149                of forgery types raises concerns...
150
151              </p>
152            </td>
153          </tr>
154
155<!---------------------------------------------ALLVB: All-in-One Long Video Understanding Benchmark------------------------------------------------------------------------------>
156        <table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;"><tbody>
157
158<!--              -->
159            <td style="padding:20px;width:25%;vertical-align:middle">
160              <div class="one">
161                <img src='images/ALLVB.jpg' height= 90%>
162              </div>
163            </td>
164            <td style="padding:20px;width:75%;vertical-align:middle">
165              <a href="https://huggingface.co/datasets/ALLVB/ALLVB">
166                <papertitle>ALLVB: All-in-One Long Video Understanding Benchmark</papertitle>
167              </a>
168              <br>
169              Xichen Tan,
170              Yuanjing Luo,
171              <strong>Yunfan Ye</strong>,
172              <a href="http://grzy.hnu.edu.cn/site/index/liufang">Fang Liu</a>,
173                <a href="http://individual.utoronto.ca/zcai/">Zhiping Cai</a>
174              <br>
175              <em>AAAI</em>, 2025
176              <br>
177              <a href="https://arxiv.org/pdf/2503.07298v2">[Paper]</a>
178              <!--
179              <a href="https://github.com/GuHuangAI/DiffusionEdge">[Code]</a>
180              -->
181              <a href="https://huggingface.co/datasets/ALLVB/ALLVB">[Dataset]</a>
182              <p></p>
183              <p style="text-align:justify">
184                From image to video understanding, the capabilities of Multimodal
185LLMs (MLLMs) are increasingly powerful. However,
186most existing video understanding benchmarks are relatively
187short, which makes them inadequate for effectively evaluating
188the long-sequence modeling capabilities of MLLMs...
189
190              </p>
191            </td>
192          </tr>
193
194<!---------------------------------------------Spatiotemporal-aware Neural Fields for Dynamic CT Reconstruction------------------------------------------------------------------------------>
195        <table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;"><tbody>
196
197<!--              -->
198            <td style="padding:20px;width:25%;vertical-align:middle">
199              <div class="one">
200                <img src='images/dynamic4DCT.jpg' height= 110%>
201              </div>
202            </td>
203            <td style="padding:20px;width:75%;vertical-align:middle">
204              <a href="https://ojs.aaai.org/index.php/AAAI/article/view/33177">
205              <a href="https://qingyangzhou69.github.io/STNF4D/">
206                <papertitle>Spatiotemporal-aware Neural Fields for Dynamic CT Reconstruction</papertitle>
207              </a>
208              <br>
209                Qingyang Zhou,
210                <strong>Yunfan Ye†</strong>,
211                <a href="http://individual.utoronto.ca/zcai/">Zhiping Cai</a>
212              <br>
213              <em>AAAI</em>, 2025
214              <br>
215              <a href="https://ojs.aaai.org/index.php/AAAI/article/view/33177">[Paper]</a>
216              <!--
217              <a href="https://arxiv.org/pdf/2401.05975.pdf">[Paper]</a>
218              <a href="https://github.com/GuHuangAI/DiffusionEdge">[Code]</a>
219              -->
220              <a href="https://qingyangzhou69.github.io/STNF4D/">[Project Page]</a>
221              <p></p>
222              <p style="text-align:justify">
223                We propose a dynamic Computed Tomography (CT) reconstruction framework called STNF4D (SpatioTemporal-aware Neural Fields). First, we represent the 4D scene using four orthogonal volumes and compress these volumes into more compact hash grids. Compared to the plane decomposition
224method, this method enhances the model’s capacity while
225keeping the representation compact and efficient...
226              </p>
227            </td>
228          </tr>
229
230
231
232<!---------------------------------------------DiffusionEdge: Diffusion Probabilistic Model for Crisp Edge Detection------------------------------------------------------------------------------>
233        <table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;"><tbody>
234          <tr bgcolor="#ffffd0">
235<!--              -->
236            <td style="padding:20px;width:25%;vertical-align:middle">
237              <div class="one">
238                <img src='images/diffusion_edge.png' height= 90%>
239              </div>
240            </td>
241            <td style="padding:20px;width:75%;vertical-align:middle">
242              <a href="https://arxiv.org/pdf/2401.02032.pdf">
243                <papertitle>DiffusionEdge: Diffusion Probabilistic Model for Crisp Edge Detection</papertitle>
244              </a>
245              <br>
246                <strong>Yunfan Ye*</strong>,
247                <a href="https://kevinkaixu.net/">Kai Xu*</a>,
248                <a href="https://github.com/GuHuangAI">Yuhang Huang†</a>,
249                <a href="https://renjiaoyi.github.io/">Renjiao Yi</a>,
250                <a href="http://individual.utoronto.ca/zcai/">Zhiping Cai</a>
251              <br>
252              <em>AAAI</em>, 2024
253              <br>
254              <a href="https://ojs.aaai.org/index.php/AAAI/article/view/28490">[Paper]</a>
255              <a href="https://github.com/GuHuangAI/DiffusionEdge">[Code]</a>
256              <a href="https://mp.weixin.qq.com/s/rHMO1nFCvz44tLVcpDWOPw">[News]</a>
257              <p></p>
258              <p style="text-align:justify">
259                Limited by the encoder-decoder architecture, learning-based edge detectors usually have difficulty predicting edge maps that satisfy both correctness and crispness. With the recent success of the diffusion probabilistic model (DPM), we found it is especially suitable for accurate and crisp edge detection...
260              </p>
261            </td>
262          </tr>
263<!---------------------------------------------NEF: Neural Edge Fields for 3D Parametric Curve Reconstruction from Multi-view Images------------------------------------------------------------------------------>
264          <tr bgcolor="#ffffd0">
265<!--              -->
266            <td style="padding:20px;width:25%;vertical-align:middle">
267              <div class="one">
268                <img src='images/NEF.png' width=112%>
269              </div>
270            </td>
271            <td style="padding:20px;width:75%;vertical-align:middle">
272              <a href="https://openaccess.thecvf.com/content/CVPR2023/papers/Ye_NEF_Neural_Edge_Fields_for_3D_Parametric_Curve_Reconstruction_From_CVPR_2023_paper.pdf">
273                <papertitle>NEF: Neural Edge Fields for 3D Parametric Curve Reconstruction from Multi-view Images</papertitle>
274              </a>
275              <br>
276                <strong>Yunfan Ye</strong>,
277                <a href="https://renjiaoyi.github.io/">Renjiao Yi</a>,
278                <a href="https://github.com/zhirui-gao">Zhirui Gao</a>,
279                <a href="https://www.zhuchenyang.net/">Chenyang Zhu</a>,
280                <a href="http://individual.utoronto.ca/zcai/">Zhiping Cai</a>,
281              <a href="https://kevinkaixu.net/">Kai Xu†</a>
282              <br>
283              <em>CVPR</em>, 2023
284              <br>
285              <a href="https://openaccess.thecvf.com/content/CVPR2023/papers/Ye_NEF_Neural_Edge_Fields_for_3D_Parametric_Curve_Reconstruction_From_CVPR_2023_paper.pdf">[Paper]</a>
286              <a href="https://github.com/yunfan1202/NEF_code">[Code]</a>
287              <a href="https://yunfan1202.github.io/NEF/">[Project Page]</a>
288              <p></p>
289              <p style="text-align:justify">
290                We study the problem of reconstructing 3D feature curves of an object from a set of calibrated multi-view images. To do so, we learn a neural implicit field representing the density distribution of 3D edges which we refer to as Neural Edge Field (NEF). Inspired by NeRF...
291              </p>
292            </td>
293          </tr>
294<!------------------------------------------------Delving into Crispness: Guided Label Refinement for Crisp Edge Detection--------------------------------------------------------------------------->
295            <tr>
296<!--              -->
297            <td style="padding:20px;width:25%;vertical-align:middle">
298              <div class="one">
299                <img src='images/crisp_edge.png' width=115%>
300              </div>
301            </td>
302            <td style="padding:20px;width:75%;vertical-align:middle">
303              <a href="https://arxiv.org/pdf/2306.15172.pdf">
304                <papertitle>Delving into Crispness: Guided Label Refinement for Crisp Edge Detection</papertitle>
305              </a>
306              <br>
307                <strong>Yunfan Ye</strong>,
308                <a href="https://renjiaoyi.github.io/">Renjiao Yi</a>,
309                <a href="https://github.com/zhirui-gao">Zhirui Gao</a>,
310                <a href="http://individual.utoronto.ca/zcai/">Zhiping Cai†</a>,
311              <a href="https://kevinkaixu.net/">Kai Xu†</a>
312              <br>
313              <em>IEEE TIP</em>, 2023
314              <br>
315              <a href="https://arxiv.org/pdf/2306.15172.pdf">[Paper]</a>
316              <a href="https://github.com/yunfan1202/Delving-into-Crispness">[Code]</a>
317
318              <p></p>
319              <p style="text-align:justify">
320                Learning-based edge detection usually suffers from predicting thick edges. Through extensive quantitative study with a new edge crispness measure, we find that noisy human-labeled edges are the main cause of thick predictions...
321
322              </p>
323            </td>
324          </tr>
325<!--------------------------------------------------------------------------------------------------------------------------->
326<!--------------------------------------------------------------------------------------------------------------------------->
327            <tr>
328            <td style="padding:20px;width:25%;vertical-align:middle">
329              <div class="one">
330                <img src='images/STEdge.png' width=115%>
331              </div>
332            </td>
333            <td style="padding:20px;width:75%;vertical-align:middle">
334              <a href="https://arxiv.org/pdf/2201.05121.pdf">
335                <papertitle>STEdge: Self-Training Edge Detection With Multilayer Teaching and Regularization</papertitle>
336              </a>
337              <br>
338                <strong>Yunfan Ye*</strong>,
339                <a href="https://renjiaoyi.github.io/">Renjiao Yi*</a>,
340                <a href="http://individual.utoronto.ca/zcai/">Zhiping Cai†</a>,
341              <a href="https://kevinkaixu.net/">Kai Xu†</a>
342              <br>
343              <em>IEEE TNNLS</em>, 2023
344              <br>
345              <a href="https://arxiv.org/pdf/2201.05121.pdf">[Paper]</a>
346              <a href="https://github.com/yunfan1202/STEdge">[Code]</a>
347              <p></p>
348              <p style="text-align:justify">
349                Learning-based edge detection has hereunto been strongly supervised with pixel-wise annotations which are tedious to obtain manually. We study the problem of self-training edge detection, leveraging the untapped wealth of large-scale unlabeled image datasets...
350              </p>
351            </td>
352          </tr>
353
354<!--------------------------------------------------------------------------------------------------------------------------->
355<!--------------------------------------------------------------------------------------------------------------------------->
356            <tr>
357            <td style="padding:20px;width:25%;vertical-align:middle">
358              <div class="one">
359                <img src='images/watermarking.jpg' width=115%>
360              </div>
361            </td>
362            <td style="padding:20px;width:75%;vertical-align:middle">
363              <a href="https://ieeexplore.ieee.org/document/10184464">
364                <papertitle>Fixing the Double Agent Vulnerability of Deep Watermarking: A Patch-Level Solution against Artwork Plagiarism</papertitle>
365              </a>
366              <br>
367                <a href="https://github.com/1024yy">Yuanjing Luo*</a>,
368                <a href="https://tongqingzhou-nudt.github.io/">
368Tongqing Zhou*</a>,
369                Shenglan Cui,
370              <strong>Yunfan Ye</strong>,
371              <a href="http://design.hnu.edu.cn/info/1023/5787.htm">Fang Liu†</a>,
372              Zhiping Cai
373              <br>
374              <em>IEEE TCSVT</em>, 2023
375              <br>
376              <a href="https://ieeexplore.ieee.org/document/10184464">[Paper]</a>
377              <a href="https://github.com/1024yy/DIPW">[Code]</a>
378              <p></p>
379              <p style="text-align:justify">
380Increasing artwork plagiarism incidents stresses the urgent need for proper copyright protection on behalf of the creators. The latest development in this context focuses on embedding watermarks via deep encoder-decoder networks...
381            </td>
382          </tr>
383<!--------------------------------------------------------------------------------------------------------------------------->
384<!---------------------------------------------------------------------------------------------------------------------------><!--------------------------------------------------------------------------------------------------------------------------->
385            <tr>
386            <td style="padding:20px;width:25%;vertical-align:middle">
387              <div class="one">
388                <img src='images/template.jpg' width=115%>
389              </div>
390            </td>
391            <td style="padding:20px;width:75%;vertical-align:middle">
392              <a href="https://arxiv.org/pdf/2303.08438.pdf">
393                <papertitle>Learning Accurate Template Matching with Differentiable Coarse-to-Fine Correspondence Refinement</papertitle>
394              </a>
395              <br>
396                <a href="https://github.com/zhirui-gao">Zhirui Gao</a>,
397                <a href="https://renjiaoyi.github.io/">Renjiao Yi</a>,
398                <a href="https://github.com/qinzheng93">Zheng Qin</a>,
399                <strong>Yunfan Ye</strong>,
400                <a href="https://www.zhuchenyang.net/">Chenyang Zhu</a>,
401                <a href="https://kevinkaixu.net/">Kai Xu†</a>
402              <br>
403              <br>
404              <em>Computational Visual Media Journal (CVMJ)</em>
405              <br>
406              <a href="https://arxiv.org/pdf/2303.08438.pdf">[Paper]</a>
407              <a href="https://github.com/zhirui-gao/Deep-Template-Matching">[Code]</a>
408              <p></p>
409              <p style="text-align:justify">
410Template matching is a fundamental task in computer vision and has been studied for decades. It plays an essential role in manufacturing industry...
411            </td>
412          </tr>
413
414<!--------------------------------------------------------------------------------------------------------------------------->
415
416<!---------------------------------------------------------------------------------------------------------------------------><!--------------------------------------------------------------------------------------------------------------------------->
417            <tr>
418            <td style="padding:20px;width:25%;vertical-align:middle">
419              <div class="one">
420                <img src='images/caption.jpg' width=115%>
421              </div>
422            </td>
423            <td style="padding:20px;width:75%;vertical-align:middle">
424              <a href="https://link.springer.com/article/10.1007/s00530-023-01178-8">
425                <papertitle>Image Captioning for Cultural Artworks: a Case Study on Ceramics</papertitle>
426              </a>
427              <br>
428              Baoying Zheng,
429              <a href="http://design.hnu.edu.cn/info/1023/5787.htm">Fang Liu†</a>,
430              Mohan Zhang,
431              <a href="https://tongqingzhou-nudt.github.io/">Tongqing Zhou</a>,
432              Shenglan Cui,
433              <strong>Yunfan Ye</strong>,
434              Yeting Guo
435              <br>
436              <br>
437              <em>Multimedia System</em>
438              <br>
439              <a href="https://link.springer.com/article/10.1007/s00530-023-01178-8">[Paper]</a>
440              <p></p>
441              <p style="text-align:justify">
442                When viewing ancient artworks, people try to build connections with them to ‘read’ the correct messages from the past. A proper descriptive caption is essential for viewers...
443            </td>
444          </tr>
445<!--------------------------------------------------------------------------------------------------------------------------->
446        <table width="100%" align="center" border="0" cellspacing="0" cellpadding="20">
447          </table>
448
449        <p align="center">Source code from <a href="https://jonbarron.info/">Jon Barron</a>'s website</p>
450        
450<script>
451          let show = true;
452          document.querySelector('#projects-show').onclick = function() {
453            if (!show) {
454              document.querySelector('#projects-box').style.height = 'auto';
455              document.querySelector('#projects-box').style.paddingBottom = '20px';
456              document.querySelector('#projects-show-text').innerHTML = 'Less';
457              show = true;
458            } else {
459              show = false;
460              document.querySelector('#projects-box').style.height = '160px';
461              document.querySelector('#projects-box').style.paddingBottom = '0px';
462              document.querySelector('#projects-show-text').innerHTML = 'More';
463            }
464          }
465        </script>
465
466</body>
467</html>

Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.