object-detection-lab / index.html
miyuki17's picture
Polish unified course paths, covers, and studio alignment
e2a0f8b verified
Raw History Blame Contribute Delete
15.3 kB
<!doctype html>
<html lang="en" data-wr92-course="det">
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1" />
<meta
name="description"
content="AI-Powered Object Detection mini-lab using COCO examples and a pretrained model"
/>
<title>AI-Powered Object Detection</title>
<link
rel="icon"
href="data:image/svg+xml,<svg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 100 100'><text y='.9em' font-size='90'>🔎</text></svg>"
/>
<link rel="stylesheet" href="./style.css" />
<link rel="stylesheet" href="./wr92-language.css">
<link rel="stylesheet" href="./wr92-theme.css">
<script defer src="./wr92-language.js"></script>
</head>
<body>
<header class="wr92-shell">
<div class="wr92-shell-inner">
<a class="wr92-brand" href="https://wr92-course-readme.static.hf.space/"><span class="wr92-brand-mark" aria-hidden="true">W·</span><span>WR92 <span class="wr92-brand-light">COURSE</span></span></a>
<nav class="wr92-site-links" aria-label="Course navigation">
<div class="wr92-nav-group">
<span class="wr92-nav-label"><span class="wr92-en">01 / LEARN</span><span class="wr92-zh">01 / 学习</span></span>
<div class="wr92-nav-items">
<a href="https://wr92-course-readme.static.hf.space/"><span class="wr92-en">Home</span><span class="wr92-zh">首页</span></a>
<a href="https://wr92-course-object-detection-lab.static.hf.space/" aria-current="page"><span class="wr92-en">Object detection</span><span class="wr92-zh">目标检测</span></a>
<a href="https://wr92-course-image-generation-lab.static.hf.space/"><span class="wr92-en">Image generation</span><span class="wr92-zh">图像生成</span></a>
</div>
</div>
<div class="wr92-nav-group">
<span class="wr92-nav-label"><span class="wr92-en">02 / LIVE TOOLS</span><span class="wr92-zh">02 / 在线工具</span></span>
<div class="wr92-nav-items">
<a href="https://huggingface.co/spaces/miyuki17/creative-technology-image-generation">SDXL Turbo</a>
<a href="https://huggingface.co/spaces/miyuki17/creative-technology-sana-sprint">SANA-Sprint</a>
</div>
</div>
<div class="wr92-nav-group wr92-nav-project">
<span class="wr92-nav-label"><span class="wr92-en">03 / APPLY</span><span class="wr92-zh">03 / 综合实践</span></span>
<div class="wr92-nav-items">
<a class="wr92-feature-link" href="https://wr92-course-object-detection-lab.static.hf.space/detect-create.html"><span class="wr92-en">Detect → Create</span><span class="wr92-zh">检测 → 生成</span></a>
</div>
</div>
</nav>
<div class="wr92-language-toolbar" role="group" aria-label="Language / 语言">
<button type="button" data-wr92-lang="en" aria-pressed="true" disabled>EN</button>
<button type="button" data-wr92-lang="zh" aria-pressed="false" disabled>中文</button>
</div>
</div>
</header>
<main class="page-shell" id="top">
<nav class="wr92-learning-path" aria-label="Learning pathway">
<span><span class="wr92-en">LEARNING PATH</span><span class="wr92-zh">学习路径</span></span>
<strong aria-current="step"><span class="wr92-en">01 · Detect</span><span class="wr92-zh">01 · 识别</span></strong>
<a href="https://wr92-course-image-generation-lab.static.hf.space/"><span class="wr92-en">02 · Generate</span><span class="wr92-zh">02 · 生成</span></a>
<a href="./detect-create.html"><span class="wr92-en">03 · Connect</span><span class="wr92-zh">03 · 串联</span></a>
</nav>
<p class="wr92-course-note"><span class="wr92-en"><strong>No Hugging Face GPU needed</strong> · Detection runs in your browser and does not use HF GPU minutes or generation calls. Device, network and model downloads can still limit it.</span><span class="wr92-zh"><strong>目标检测不消耗 Hugging Face GPU</strong> · 检测在本机浏览器完成,没有 ZeroGPU 分钟或生成次数限制,但仍受设备、网络和模型下载影响。</span> <a href="https://huggingface.co/docs/hub/spaces-zerogpu" target="_blank" rel="noopener noreferrer"><span class="wr92-en">ZeroGPU policy →</span><span class="wr92-zh">ZeroGPU 官方规则 →</span></a></p>
<nav class="guide-linkbar" aria-label="Course guide">
<span>Building your own AI mini app?</span>
<a href="./how-to-start.html">How to Start <span aria-hidden="true">→</span></a>
<a href="https://wr92-course-image-generation-lab.static.hf.space/" target="_blank" rel="noopener noreferrer">Image Generation <span aria-hidden="true">→</span></a>
</nav>
<!-- =========================================================
HERO
========================================================== -->
<header class="hero wr92-detection-hero">
<div class="wr92-lesson-copy">
<p class="eyebrow">01 / <span class="wr92-en">LEARN · IMAGE UNDERSTANDING</span><span class="wr92-zh">学习 · 图像理解</span></p>
<h1>
AI-Powered<br />
Object Detection
</h1>
<p class="lede">
Pick a COCO example or upload your own image.
Then choose a pretrained detector and let AI answer:
<strong>
what objects are here, and where are they?
</strong>
</p>
<div
class="lesson-strip"
aria-label="Mini-lab overview"
>
<span>
<b>Task</b>
Object detection
</span>
<span>
<b>Data</b>
COCO validation examples
</span>
<span>
<b>Models</b>
RT-DETR · DETR · YOLOS Tiny
</span>
</div>
</div>
<figure class="wr92-lesson-visual">
<img src="./examples/000000189806.jpg" alt="Real COCO example input showing a cat and dog indoors, before running object detection" width="640" height="425" fetchpriority="high">
<figcaption><strong><span class="wr92-en">REAL IMAGE INPUT</span><span class="wr92-zh">真实输入图片</span></strong><span><span class="wr92-en">COCO val2017 · 000000189806 · not model output</span><span class="wr92-zh">COCO val2017 · 000000189806 · 不是模型预测图</span></span></figcaption>
</figure>
</header>
<!-- =========================================================
MAIN WORKSPACE
========================================================== -->
<section
class="workspace"
aria-label="Object detection mini-lab"
>
<!-- =======================================================
LEFT: CONTROLS
======================================================== -->
<div class="controls card">
<!-- STEP 1 -->
<div class="step">
<span class="step-number">
1
</span>
<div>
<p class="control-label sample-title">
Choose an image
</p>
<p class="helper">
Start with a tested COCO example,
or try a photo of your own.
</p>
</div>
</div>
<!-- COCO EXAMPLES -->
<div class="sample-head coco-head">
<div>
<p class="input-kicker">
Start here · COCO examples
</p>
<p class="helper">
Four images are drawn at random
from 100 bundled COCO validation examples.
</p>
</div>
<button
id="shuffle-examples"
class="shuffle-button"
type="button"
>
Shuffle
</button>
</div>
<div
id="sample-grid"
class="sample-grid"
aria-label="COCO example images"
></div>
<!-- OR -->
<div class="or-row">
<span>
or try your own
</span>
</div>
<!-- IMAGE UPLOAD -->
<label
class="control-label"
for="image-input"
>
Upload an image
</label>
<p class="helper">
Useful for seeing where the model
generalizes well—and where it does not.
</p>
<input
id="image-input"
class="file-input"
type="file"
accept="image/*"
/>
<div class="divider"></div>
<!-- STEP 2 -->
<div class="step">
<span class="step-number">
2
</span>
<div class="grow">
<label
class="control-label"
for="model-select"
>
Choose detector
</label>
<select
id="model-select"
class="model-select"
></select>
<p
id="model-note"
class="helper"
>
RT-DETR is the recommended classroom default.
</p>
</div>
</div>
<div class="divider"></div>
<!-- STEP 3 -->
<div class="step">
<span class="step-number">
3
</span>
<div class="grow">
<div class="slider-row">
<label
class="control-label"
for="threshold"
>
Confidence threshold
</label>
<output
id="threshold-value"
for="threshold"
>
0.50
</output>
</div>
<input
id="threshold"
type="range"
min="0.10"
max="0.95"
value="0.50"
step="0.05"
/>
<p class="helper">
Higher values only keep predictions
the model is more confident about.
</p>
</div>
</div>
<!-- RUN BUTTON -->
<button
id="run-button"
class="run-button"
type="button"
disabled
>
Run AI Detection
</button>
<!-- STATUS -->
<p
id="status"
class="status"
aria-live="polite"
>
Choose a COCO example or upload your own image.
</p>
<!-- PROGRESS -->
<div
id="progress-wrap"
class="progress-wrap"
hidden
>
<div
id="progress-bar"
class="progress-bar"
></div>
</div>
</div>
<!-- =======================================================
RIGHT: RESULT
======================================================== -->
<div class="result card">
<div class="result-head">
<div>
<p class="section-kicker">
AI prediction
</p>
<h2>
What did the detector find?
</h2>
</div>
<span
id="runtime-badge"
class="badge"
>
Waiting
</span>
</div>
<!-- RESULT CANVAS -->
<div
id="canvas-wrap"
class="canvas-wrap empty"
>
<canvas
id="result-canvas"
aria-label="Detection result"
></canvas>
<p id="canvas-placeholder">
Choose an image first.
The original image appears here before detection.
</p>
</div>
<!-- SUMMARY -->
<div
id="summary"
class="summary"
>
No detections yet.
</div>
<!-- DETECTION TABLE -->
<div class="table-wrap">
<table>
<thead>
<tr>
<th>Object</th>
<th>Confidence</th>
<th>Bounding box</th>
</tr>
</thead>
<tbody id="detections-body">
<tr class="empty-row">
<td colspan="3">
Run the detector to see details.
</td>
</tr>
</tbody>
</table>
</div>
</div>
</section>
<aside class="wr92-next-step" id="detect-create" aria-labelledby="wr92-next-step-heading">
<div>
<p class="wr92-flow-kicker">03 / <span class="wr92-en">PUT THE LESSONS TOGETHER</span><span class="wr92-zh">综合运用两门课程</span></p>
<h2 id="wr92-next-step-heading"><span class="wr92-en">What can you make from these labels?</span><span class="wr92-zh">用检测到的物体,能创作什么新图片?</span></h2>
<p><span class="wr92-en">Take object detection into the dedicated three-step studio: pick a photo, review its detected concepts, then generate a fresh image without leaving that workspace.</span><span class="wr92-zh">在独立的三步工作台中,选照片、查看识别出的概念,然后直接生成一张新图,无需来回切换页面。</span></p>
</div>
<a class="wr92-next-link" href="./detect-create.html"><span class="wr92-en">Open connected studio</span><span class="wr92-zh">进入一条龙实验室</span><span aria-hidden="true">→</span></a>
</aside>
<!-- =========================================================
EXPLANATION
========================================================== -->
<section class="explain card">
<p class="section-kicker">
What you are doing
</p>
<h2 class="explain-title">
Use a learned AI capability instead of rebuilding
the traditional task from scratch.
</h2>
<p class="explain-copy">
COCO contains everyday-object images with annotations.
The selected pretrained detector has already learned
the object-detection task from labeled data.
In this mini-lab, you can switch models and compare
how their predictions differ on the same image.
</p>
<div
class="flow"
aria-label="Object detection flow"
>
<span>
COCO example or your image
</span>
<span class="arrow">
→
</span>
<span>
Selected pretrained detector
</span>
<span class="arrow">
→
</span>
<span>
Boxes + labels + confidence
</span>
</div>
<p class="extension-note">
<strong>
Compare models:
</strong>
keep the image fixed,
switch detector,
and observe how predictions,
confidence,
and speed change.
</p>
</section>
<!-- =========================================================
FOOTER
========================================================== -->
<footer>
Hosted as a Hugging Face Static Space ·
Powered by Transformers.js
</footer>
</main>
<!-- ===========================================================
APP
============================================================ -->
<script
type="module"
src="./app.js"
></script>
</body>
</html>