Yidun (易盾) is the CAPTCHA and risk-control platform from NetEase, used by many Chinese sites to protect login, sign-up and comment forms. Out of the box it shows a slider puzzle, but depending on the risk score a site can switch to icon clicks, word selection or a plain checkbox.
How to identifyLink to How to identify
- A hidden
NECaptchaValidatefield holding a Base64 value - Scripts and images from
cstaticdun.126.netordun.163.com - Classes such as
yidun,yidun_slider,yidun_bg-imgandyidun_jigsaw - The prompt "请向右滑动完成拼图" ("Slide to complete the puzzle")
- A
captchaId, a 32-character hex string passed toinitNECaptcha - Verification traffic to
c.dun.163.com
Common challenge typesLink to Common challenge types
- Slide: fit the jigsaw piece into the gap.
- Click: tap characters or icons in the order asked.
- Icon / word: choose the matching symbols in a grid.
- Checkbox: low-risk sessions may get through with one click.
How it worksLink to How it works
- The page calls
initNECaptchawith itscaptchaIdand displays the widget. - Once the challenge is solved, the SDK puts a Base64 token into
NECaptchaValidate. - The site's backend sends that token to Yidun's verification API (
c.dun.163.com/api/v2/verify) with thecaptchaId, a user identifier and a signed request (secretId, timestamp, nonce and signature). - A valid token returns a pass. It can be used once and expires after roughly 20 minutes.
- Because the token is tied to the challenge, and usually to the session, it has to be submitted from the same client.
Free & open-source bypass methodsLink to Free & open-source bypass methods
| Tool | What it does | Success Rate | Link |
|---|---|---|---|
| OpenCV template matching | Finds the gap with matchTemplate or edges, then drags |
70–90% | opencv/opencv-python |
| PuzzleSolver | A reusable slider solver with natural trajectories | 70–90% | vsmutok/Puzzle-Captcha-Solver |
| ddddocr | Spots slider gaps and simple click targets | 65–85% | sml2h3/ddddocr |
| YOLOv8 | A detector you train for icon and word click tasks | 80–95% (trained) | ultralytics/ultralytics |
| DrissionPage / Playwright | Dependable widget automation and token extraction | 70–90% | g1879/DrissionPage · microsoft/playwright-python |
Practical approachLink to Practical approach
- Load the page in a real browser and wait for the Yidun widget.
- Save the background and the piece as two separate images, not as a screenshot that rescales them.
- Find the gap's x-coordinate with Canny edges and
matchTemplate. - Drag along an eased path with a little jitter. An instant straight drag gets rejected.
- When the success state shows, read
NECaptchaValidateand send the form. - If the widget escalates to click or icon mode, bring in an object detector or start over with a fresh session.
Practical tipsLink to Practical tips
- The background and piece usually arrive as separate files. Use the originals so the dimensions stay right.
- Yidun examines the whole drag path, not only where the piece ends up.
- A failure typically triggers a new challenge. Vary the trajectory instead of replaying the old one.
NECaptchaValidateis Base64 and may need URL-encoding in the request.- Align the browser locale and IP with the target. Chinese sites tend to score foreign sessions harshly.