Repository navigation
Expand file tree
/
Copy pathindex.html
More file actions
511 lines (484 loc) · 24.1 KB
/
Copy pathindex.html
File metadata and controls
511 lines (484 loc) · 24.1 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
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<meta name="author" content="DingTian Xue" />
<meta name="description" content="Academic homepage of DingTian Xue (薛丁天) — Computer Vision Researcher, edge AI deployment, lightweight neural networks, hyperspectral image clustering, Chinese NER, sequence labeling." />
<meta name="keywords" content="DingTian Xue, 薛丁天, computer vision, edge AI, hyperspectral, deep learning, MobileNet, S2DL, Chinese NER, sequence labeling, BERT, CRF, LoRA, Qwen" />
<title>DingTian Xue | Computer Vision Researcher</title>
<style>
:root {
--bg: #0f1216;
--card: rgba(18, 21, 25, 0.72);
--card-border: rgba(200, 210, 220, 0.14);
--ink: #f1f4f7;
--muted: #9aa6b3;
--accent: #9fc3e8;
--accent-2: #7d8ea0;
--line: rgba(210, 220, 230, 0.1);
}
* { box-sizing: border-box; margin: 0; padding: 0; }
html { scroll-behavior: smooth; }
/* perf-optimised grey brick-wall background. Keeps the chessboard breathing
via SVG-internal <animate> only (GPU-cheap). No backdrop-filter / blend-mode /
animated filter / large shadow. The 3.png atmosphere and grain layers were
removed per request — only the brick wall (a single SVG) provides the visual. */
body {
font-family: "Segoe UI", "Source Sans 3", Roboto, "Helvetica Neue", Arial, "PingFang SC", "Microsoft YaHei", sans-serif;
background: var(--bg);
color: var(--ink);
line-height: 1.7;
min-height: 100vh;
overflow-x: hidden;
}
/* The brick wall is the only background layer. Its chessboard opacity animation
lives inside bricks.svg (<animate dur="9s">). No CSS filter, no transform anim. */
.bricks {
position: fixed;
inset: 0;
z-index: -1;
background-image: url("bricks.svg");
background-size: 160px 160px;
background-repeat: repeat;
opacity: 0.45;
}
/* NOTE: .bg (3.png) and .grain (noise SVG) layers removed per request. */
.wrap { max-width: 880px; margin: 0 auto; padding: 0 22px; position: relative; z-index: 1; }
/* ---------- header ---------- */
header { padding: 72px 0 36px; text-align: center; }
.avatar {
width: 148px;
height: 148px;
border-radius: 50%;
object-fit: cover;
border: 2px solid rgba(255, 255, 255, 0.3);
margin-bottom: 22px;
}
header h1 {
font-size: 2.15rem;
font-weight: 300;
letter-spacing: 2px;
}
header h1 .cn { color: var(--muted); font-weight: 300; font-size: 1.5rem; letter-spacing: 1px; }
header .role {
margin-top: 8px;
font-weight: 500;
text-transform: uppercase;
font-size: 0.82rem;
letter-spacing: 3px;
background: linear-gradient(90deg, var(--accent), var(--accent-2));
-webkit-background-clip: text;
background-clip: text;
color: transparent;
}
header .tagline { margin-top: 12px; color: var(--muted); font-size: 0.97rem; font-weight: 300; }
.contact {
margin-top: 20px;
display: flex;
flex-wrap: wrap;
justify-content: center;
gap: 6px 20px;
font-size: 0.9rem;
}
.contact a {
color: var(--ink);
text-decoration: none;
border-bottom: 1px dotted rgba(255, 255, 255, 0.35);
padding-bottom: 1px;
transition: color 0.25s, border-color 0.25s;
}
.contact a:hover { color: var(--accent); border-color: var(--accent); }
/* ---------- nav ---------- */
nav {
position: sticky; top: 0;
z-index: 10;
padding: 12px 0;
background: rgba(15, 18, 22, 0.85);
border-bottom: 1px solid var(--line);
}
nav .wrap { display: flex; justify-content: center; flex-wrap: wrap; gap: 4px 26px; }
nav a {
color: var(--muted);
text-decoration: none;
font-size: 0.88rem;
font-weight: 400;
letter-spacing: 0.6px;
padding: 3px 2px;
transition: color 0.25s;
}
nav a:hover { color: var(--accent); }
/* ---------- sections ---------- */
main { padding: 36px 0 12px; }
section {
background: rgba(18, 21, 25, 0.82);
border: 1px solid var(--card-border);
border-radius: 16px;
padding: 30px 32px;
margin-bottom: 22px;
}
h2 {
font-size: 1.22rem;
font-weight: 500;
letter-spacing: 1px;
margin-bottom: 16px;
padding-bottom: 10px;
position: relative;
display: inline-block;
}
h2::after {
content: "";
position: absolute;
left: 0; bottom: 0;
width: 100%;
height: 2px;
background: linear-gradient(90deg, var(--accent), transparent);
border-radius: 2px;
}
h3 { font-size: 1rem; font-weight: 500; margin: 16px 0 8px; color: #dfe5ea; }
p { margin-bottom: 10px; font-weight: 300; }
ul { margin: 6px 0 10px 22px; }
li { margin-bottom: 7px; font-weight: 300; }
li::marker { color: var(--accent); }
.muted { color: var(--muted); }
.small { font-size: 0.88rem; }
a { color: var(--accent); text-decoration: none; transition: opacity 0.25s; }
a:hover { text-decoration: underline; }
code {
background: rgba(255, 255, 255, 0.08);
padding: 1px 7px;
border-radius: 5px;
font-size: 0.86em;
border: 1px solid rgba(255, 255, 255, 0.06);
}
.item { margin-bottom: 20px; }
.item:last-child { margin-bottom: 0; }
.item-head { display: flex; justify-content: space-between; align-items: baseline; flex-wrap: wrap; gap: 4px 14px; }
.item-head .title { font-weight: 500; font-size: 1rem; color: #f2f5f8; }
.item-head .date { color: var(--muted); font-size: 0.83rem; white-space: nowrap; }
.org { color: var(--muted); font-size: 0.9rem; margin-bottom: 6px; }
.badge {
display: inline-block;
font-size: 0.68rem;
font-weight: 600;
letter-spacing: 0.5px;
padding: 1px 10px;
border-radius: 999px;
margin-left: 8px;
vertical-align: 2px;
background: rgba(159, 195, 232, 0.12);
color: var(--accent);
border: 1px solid rgba(159, 195, 232, 0.3);
}
.tags { margin-top: 8px; }
.tag {
display: inline-block;
background: rgba(255, 255, 255, 0.07);
color: #d6dde4;
font-size: 0.78rem;
font-weight: 300;
padding: 3px 12px;
border-radius: 999px;
margin: 3px 5px 3px 0;
border: 1px solid rgba(255, 255, 255, 0.08);
transition: border-color 0.25s, color 0.25s;
}
.tag:hover { border-color: var(--accent); color: var(--accent); }
.feature {
background: linear-gradient(135deg, rgba(159, 195, 232, 0.1), rgba(125, 142, 160, 0.12));
border: 1px solid rgba(159, 195, 232, 0.24);
border-radius: 12px;
padding: 16px 18px;
margin: 14px 0 4px;
font-weight: 300;
}
/* ---------- news & publications ---------- */
.news-date {
display: inline-block;
min-width: 56px;
padding: 1px 10px;
margin-right: 8px;
border-radius: 999px;
background: rgba(255, 255, 255, 0.06);
border: 1px solid rgba(255, 255, 255, 0.1);
color: var(--accent);
font-weight: 600;
font-size: 0.78rem;
letter-spacing: 0.5px;
}
#news li { line-height: 1.85; }
.pub-item {
padding: 12px 0 14px;
border-bottom: 1px dashed rgba(255, 255, 255, 0.08);
}
.pub-item:first-of-type { padding-top: 4px; }
.pub-item:last-of-type { border-bottom: 0; padding-bottom: 0; }
.pub-line { display: flex; flex-wrap: wrap; gap: 6px 10px; align-items: baseline; margin-bottom: 4px; }
.pub-tag {
display: inline-block;
font-size: 0.7rem;
font-weight: 600;
letter-spacing: 0.5px;
padding: 1px 9px;
border-radius: 999px;
background: rgba(159, 195, 232, 0.14);
color: var(--accent);
border: 1px solid rgba(159, 195, 232, 0.3);
}
.pub-venue {
color: var(--muted);
font-size: 0.83rem;
font-weight: 400;
}
.pub-title {
font-weight: 500;
color: #f2f5f8;
font-size: 0.97rem;
margin-bottom: 4px;
}
.pub-authors { color: var(--muted); font-size: 0.85rem; margin-bottom: 2px; }
.pub-item p { margin-top: 6px; color: var(--muted); }
.muted { color: var(--muted); }
/* ---------- footer ---------- */
footer { text-align: center; color: var(--muted); font-size: 0.83rem; padding: 30px 0 46px; font-weight: 300; }
footer a { color: var(--muted); }
footer a:hover { color: var(--accent); }
@media (max-width: 600px) {
header { padding: 48px 0 26px; }
section { padding: 22px 20px; }
.bg { background-size: cover; }
}
</style>
</head>
<body>
<div class="bricks" aria-hidden="true"></div>
<header>
<div class="wrap">
<img class="avatar" src="1.jpg" alt="Portrait of DingTian Xue" />
<h1>DingTian Xue <span class="cn">薛丁天</span></h1>
<div class="role">Computer Vision Researcher · Edge AI & Mobile Vision</div>
<div class="tagline">Lightweight models · Efficient deployment · Hyperspectral analysis · Limited-supervision learning</div>
<div class="contact">
<a href="mailto:zhongliwansui@outlook.com">✉ zh****wansui@outlook.com</a>
<a href="https://github.com/KBber" target="_blank" rel="noopener">GitHub</a>
<a href="https://www.linkedin.com/in/dingtian-xue" target="_blank" rel="noopener">LinkedIn</a>
<a href="https://kbber.github.io" target="_blank" rel="noopener">kbber.github.io</a>
</div>
</div>
</header>
<nav>
<div class="wrap">
<a href="#about">About</a>
<a href="#news">News</a>
<a href="#research">Research</a>
<a href="#publications">Publications</a>
<a href="#projects">Projects</a>
<a href="#experience">Experience</a>
<a href="#education">Education</a>
<a href="#honors">Honors & Certificates</a>
<a href="#skills">Skills</a>
</div>
</nav>
<main class="wrap">
<section id="about">
<h2>👤 About Me</h2>
<p>
I am a Computer Vision Researcher with a B.E. in Computer Science from
<strong>Yanshan University</strong> (2025, GPA 80.36/100, Top 33%), currently working as a
Computer Vision Engineer at Tiandy Technologies while pursuing self-directed research toward
Ph.D. opportunities. My research interests lie at the intersection of
<strong>edge AI deployment, lightweight neural networks, and hyperspectral image analysis</strong>
— building models that are accurate under limited supervision and efficient enough to run on real devices.
</p>
<p>
I am currently <strong>independently reproducing and extending the S2DL method</strong>
(Cui et al., IEEE TGRS 2024) for unsupervised hyperspectral image clustering — a line of work I plan
to continue at the doctoral level. My undergraduate edge-AI project, a 245-class mobile garbage
classifier, is fully open-sourced on GitHub.
</p>
<div class="feature">
<strong>Looking for:</strong> Ph.D. / Research Assistant positions in computer vision, with a focus on
annotation-efficient learning, hyperspectral and remote-sensing analysis, and edge deployment.
See my <a href="#research">research proposal</a> outline below.
</div>
</section>
<section id="news">
<h2>🔥 News & Updates</h2>
<ul>
<li><span class="news-date">2026</span> · Released <strong><a href="https://github.com/KBber/chinese-ner-sequence-labeling" target="_blank" rel="noopener">chinese-ner-sequence-labeling</a></strong> — a four-paradigm study (BERT+Linear, BERT+CRF, LLM API, LLM SFT-LoRA) of Chinese NER on CLUENER2020 & People's Daily.</li>
<li><span class="news-date">2026</span> · Building this site and self-directedly pushing toward a Ph.D. in computer vision; actively extending the S2DL reproduction to Salinas / SalinasA and UAV imagery.</li>
<li><span class="news-date">2025</span> · Joined Tiandy Technologies as a Computer Vision Engineer (presales & on-site delivery); continued the S2DL reproduction in MATLAB R2024b on the side.</li>
<li><span class="news-date">2025</span> · Graduated from Yanshan University (B.E. in Computer Science, Top 33%) and open-sourced the 245-class edge-AI garbage classifier.</li>
</ul>
</section>
<section id="research">
<h2>🧪 Research Interests</h2>
<ul>
<li><strong>Edge AI & lightweight neural networks</strong> — model compression, quantization, mobile/embedded deployment</li>
<li><strong>Hyperspectral image analysis</strong> — unsupervised clustering, diffusion geometry, superpixel priors</li>
<li><strong>Limited-supervision learning</strong> — self-supervised and semi-supervised methods for scarce labels</li>
<li><strong>Small-object recognition</strong> — detection/segmentation of tiny, low-contrast targets in remote sensing and medical imagery</li>
</ul>
<h3>Doctoral Research Direction (Proposal Outline)</h3>
<p class="small">
<em>Annotation-Efficient and Edge-Deployable Learning for Hyperspectral Clustering and Small-Object
Recognition in Remote Sensing and Medical Imaging.</em>
Building directly on my reproduction of S2DL, the proposed work investigates:
(1) geometry-aware, label-efficient hyperspectral clustering; (2) small-object detection/segmentation
under limited annotations; and (3) deployment-aware compression that preserves small and subtle targets.
Target venues: TGRS, MICCAI, ISBI, ICMR, IGARSS.
</p>
</section>
<section id="publications">
<h2>📄 Publications & Outputs</h2>
<div class="pub-item">
<div class="pub-line"><span class="pub-tag">Technical Note</span><span class="pub-venue">Independent MATLAB Reproduction</span></div>
<div class="pub-title">Superpixel-based and Spatially-regularized Diffusion Learning for Unsupervised Hyperspectral Image Clustering (S2DL)</div>
<div class="pub-authors"><strong>DingTian XUE</strong> (reproducing)</div>
<div class="pub-venue">Original: K. Cui et al., <em>IEEE TGRS</em>, 2024 · <code>github.com/KBber/S2DL</code></div>
<p class="small">Indian Pines: <strong>OA = 64.70%, Kappa = 0.6020</strong> (matches published). Extension to Salinas/SalinasA and UAV imagery in progress.</p>
</div>
<div class="pub-item">
<div class="pub-line"><span class="pub-tag">Open Source</span><span class="pub-venue">Course Capstone</span></div>
<div class="pub-title">chinese-ner-sequence-labeling — Four-Paradigm Comparison of Chinese NER</div>
<div class="pub-authors"><strong>DingTian XUE</strong></div>
<div class="pub-venue"><code>github.com/KBber/chinese-ner-sequence-labeling</code> · MIT license</div>
<p class="small">BERT + Linear / BERT + CRF / LLM API / LLM SFT-LoRA · CLUENER2020 & People's Daily · seqeval entity-level F1.</p>
</div>
<div class="pub-item">
<div class="pub-line"><span class="pub-tag">Open Source</span><span class="pub-venue">Undergraduate Project</span></div>
<div class="pub-title">edgeai-trash-classifier — 245-class Mobile Garbage Classification</div>
<div class="pub-authors"><strong>DingTian XUE</strong></div>
<div class="pub-venue"><code>github.com/KBber/edgeai-trash-classifier</code> · MIT license</div>
<p class="small">MobileNetV2 + 3-layer MLP head · 16.2 MB · ~92% Top-1 · ~65 ms end-to-end on a mid-tier Snapdragon device.</p>
</div>
<p class="small muted">Papers and theses under the supervision of Prof. Kangning Cui are in preparation; see <a href="#research">research</a> for the proposed doctoral direction.</p>
</section>
<section id="projects">
<h2>💻 Projects & Open Source</h2>
<div class="item">
<div class="item-head">
<span class="title">chinese-ner-sequence-labeling — Four-Paradigm Chinese NER Study <span class="badge">Open Source</span></span>
<span class="date">2026</span>
</div>
<div class="org">NLP · Sequence Labeling · <code>github.com/KBber/chinese-ner-sequence-labeling</code> (MIT)</div>
<ul>
<li>A systematic comparison of <strong>four NER paradigms</strong> on Chinese text: <strong>BERT + Linear</strong> (per-token classification), <strong>BERT + CRF</strong> (global sequence decoding), <strong>LLM API prompting</strong> (Qwen-Plus zero-/few-shot), and <strong>LLM SFT-LoRA</strong> (Qwen2-0.5B instruction-tuned).</li>
<li>Built on <strong>CLUENER2020</strong> (10 fine-grained entity types, 10,748 / 1,343 train/val) and the <strong>People's Daily</strong> classical NER benchmark (PER / ORG / LOC).</li>
<li>Implemented the full pipeline: BIO conversion with subword alignment, AdamW with layer-wise learning rates & linear warmup, seqeval entity-level F1, LoRA adapter training with loss masking, and a unified <code>compare_results.py</code> for cross-paradigm aggregation.</li>
<li>Designed as a <strong>Week-7 course capstone</strong> for <em>Sequence Labeling</em>; each paradigm answers a specific pedagogical question — when global decoding matters, how far prompted LLMs go on structured prediction, and the engineering trade-off of generative vs. discriminative NER.</li>
</ul>
</div>
<div class="item">
<div class="item-head">
<span class="title">Independent Reproduction of S2DL <span class="badge">Research</span></span>
<span class="date">Aug. 2025 – Present</span>
</div>
<div class="org">"Superpixel-based and Spatially-regularized Diffusion Learning for Unsupervised Hyperspectral Image Clustering" — Cui et al., IEEE TGRS 2024 · <code>github.com/KBber/S2DL</code></div>
<ul>
<li>Independently re-implemented S2DL in <strong>MATLAB R2024b</strong>; fixed the MinGW/MEX toolchain and modernized syntax (4 commits to the official fork).</li>
<li>Reproduced the published result on Indian Pines: <strong>OA = 64.70%, Kappa = 0.6020</strong>.</li>
<li>Built a single-evaluation entry-point script that bypasses the 225k-eval grid search for fast hyper-parameter iteration.</li>
<li><em>Next:</em> extend to Salinas/SalinasA and UAV imagery; rebuild the S2DL vs. SRDL/D-VIC/LUND comparison table.</li>
</ul>
</div>
<div class="item">
<div class="item-head">
<span class="title">edgeai-trash-classifier — 245-class Edge AI Garbage Classification <span class="badge">Open Source</span></span>
<span class="date">Jan. 2024 – Apr. 2025</span>
</div>
<div class="org">Graduation Project · <code>github.com/KBber/edgeai-trash-classifier</code> (MIT)</div>
<ul>
<li>Compared MobileNetV2 / ResNet / plain CNN backbones with ImageNet transfer learning; best model reaches <strong>~92% Top-1</strong> on a held-out 245-class set at <strong>16.2 MB</strong>.</li>
<li>Deployed with PyTorch-Mobile (TorchScript + <code>optimize_for_mobile</code>) into an Android app (CameraX live detection); end-to-end latency <strong>~65 ms</strong> on a mid-tier Snapdragon device.</li>
<li>Released with deterministic seeds, smoke tests and GitHub-Actions CI (reproducible-research hygiene).</li>
</ul>
</div>
<div class="item">
<div class="item-head">
<span class="title">Vision-based Temperature-Measuring Patrol Vehicle <span class="badge">Project</span></span>
<span class="date">Jul. 2023 – Jul. 2024</span>
</div>
<div class="org">Project Lead (team of 3) · Raspberry Pi + STM32</div>
<ul>
<li>Built a low-cost line-following and temperature-sensing inspection robot (Raspberry Pi camera + STM32 motor controller).</li>
<li>Implemented a classical-CV pipeline: acquisition → ROI selection → edge-/Hough-based trajectory detection → direction decision → actuator command; stable line-following under variable lighting with a mobile remote-viewing interface.</li>
</ul>
</div>
</section>
<section id="experience">
<h2>💼 Work Experience</h2>
<div class="item">
<div class="item-head">
<span class="title">Computer Vision Engineer — Presales & On-site Delivery</span>
<span class="date">Jul. 2025 – Present</span>
</div>
<div class="org">Tiandy Technologies Co., Ltd. · China</div>
<ul>
<li>Embedded deep-learning vision models (object detection / segmentation) into product pipelines for the political-legal and smart-city verticals.</li>
<li>Coordinated dataset curation and pre-processing of unstructured on-site imagery; converted raw footage into model-ready training sets.</li>
<li>Migrated PyTorch/ONNX models to GPU/edge servers and tuned inference for real-time constraints (FPS, latency); resolved deployment-environment issues.</li>
</ul>
</div>
</section>
<section id="education">
<h2>🎓 Education</h2>
<div class="item">
<div class="item-head">
<span class="title">B.E. in Computer Science and Technology</span>
<span class="date">Sep. 2021 – Jun. 2025</span>
</div>
<div class="org">Yanshan University · Qinhuangdao, Hebei, China</div>
<p class="small">GPA-P 80.36 / 100 (Top 33%). Selected coursework: Discrete Mathematics 94, Computer Networks 90, Linear Algebra 89, Computer Control Theory 89, Computer Organisation Project 90, Compiler Principles 89, Object-Oriented Programming 87, Software Engineering 82.</p>
</div>
<div class="item">
<div class="item-head">
<span class="title">M.S. in Computer Technology (085404) — Admitted, declined</span>
<span class="date">Admitted 2025</span>
</div>
<div class="org">Dalian University of Technology</div>
<p class="small">Admitted via the national graduate entrance examination; declined enrollment to pursue a full-time industry role and self-directed research, transitioning toward a research-aligned Ph.D./RA trajectory.</p>
</div>
</section>
<section id="honors">
<h2>🏆 Honors & Certificates</h2>
<div class="item-head">
<span class="title">Honors</span>
</div>
<ul>
<li>Second Prize — Blue Bridge Cup Software Competition (Provincial), 2024</li>
<li>Second & Third-Class Scholarship — Yanshan University, 2021–2023</li>
<li>Outstanding Student Cadre — Yanshan University, 2022</li>
<li>Excellence Award — "Internet+" Innovation & Entrepreneurship Competition (University Level), 2022</li>
<li>University-level Project — College Students' Innovation & Entrepreneurship Training Program, 2022–2024</li>
<li>Outstanding Individual in Production Internship — Yanshan University, 2024</li>
</ul>
<div class="item-head" style="margin-top: 16px;">
<span class="title">Certificates</span>
</div>
<ul>
<li>College English Test Band 6 (CET-6), 458/710 — 2023</li>
<li>National Computer Rank Examination — Level 4 — 2024</li>
<li>Mandarin Proficiency Test — Level 2A — 2022</li>
</ul>
</section>
<section id="skills">
<h2>🛠️ Technical Skills</h2>
<div class="tags">
<span class="tag">Python</span><span class="tag">PyTorch</span><span class="tag">MATLAB</span><span class="tag">C/C++</span>
<span class="tag">TensorFlow/Keras</span><span class="tag">PyTorch Mobile</span><span class="tag">OpenCV</span><span class="tag">YOLO</span>
<span class="tag">MobileNetV2 / ResNet</span><span class="tag">Android (CameraX)</span><span class="tag">ONNX</span><span class="tag">Git / CI</span>
<span class="tag">HuggingFace Transformers</span><span class="tag">PEFT / LoRA</span><span class="tag">BERT</span><span class="tag">pytorch-crf</span>
<span class="tag">seqeval</span><span class="tag">Qwen / LLM API</span>
</div>
<h3>Research skills (self-studying)</h3>
<p class="small">Spectral methods, manifold learning, graph-based clustering, diffusion learning; NumPy, scikit-learn, PyTorch Geometric; NLP / sequence labeling (BERT, CRF, LoRA SFT); LaTeX for academic writing.</p>
</section>
</main>
<footer>
<div class="wrap">
© 2026 DingTian Xue · Built with plain HTML/CSS · <a href="https://github.com/KBber/KBber.github.io" target="_blank" rel="noopener">Source</a>
</div>
</footer>
</body>
</html>