En / 中文


creating a continuous surveillance system to help find missing children

Face Recognition Algorithm

The machine learning algorithm extracts 128 embeddings from user submitted photos. Zhongxun identifies similar photos by calculating the Euclidean distance between their embeddings.

Social Media and User Feedback

Zhongxun continuously scans photos of missing children and posts similar photos to social media. User feedback helps improve its searching algorithms.

A Continuous Search Mechanism

Aggregate scattered signals over time. Keep tracking of every possible match.

How to Use

Zhongxun's Weibo:人脸识别寻找被拐儿童

  • Snap a Photo

    Send to Zhongxun via Weixin or Weibo. For better accuracy, please include only frontal view of the person in your photo while we are working on more robust face detection methods.

  • Compare Similar Faces

    Based on face comparison results, Zhongxun will return a list of best-matching photos of missing children.

  • Give Feedback

    Your feedback (reply: similar / not similar) will help improve Zhongxun's search algorithm

  • Follow Zhongxun's Weibo

    Follow Zhongxun's Weibo:人脸识别寻找被拐儿童. Zhongxun’s search algorithm will continuously post similar photos to Weibo.


Believe in Individual and Technology

Haiyun Lu

Machine Learning Scientist

Hang Guo



Sharing our labeled dataset

We labeled over 10 thousands Chinese faces with age and gender information. Sharing this result with people who are also in this effort.

Made meaningful progress in age progression face prediction!

Now with more confidence in predicting how a missing child will look years later, or predicting childhood facial images from an adult.

Zhongxun.life Received Good Design Award!

Feeling humbled and grateful, many thanks to GDA juries and our supporters. We will continue the good work.

Experimenting GAN-based Age Progression Prediction

First trial on predicting face progression prediction with GAN (Generative Advisory Network) based machine learning model. A long way to go, but the future is Here!

Presenting Zhongxun @ Good Design Award Second Screening

We are at Hong Kong to present Zhongxun to juries of this year's Good Design Award. Lots of learning and excitements!

Now Zhongxun will return photos directly to Weibo Chat window!

We learned from our users, a few times at least, Weibo’s spam filter mistakenly thought zhongxun's site has fraud risk. This is because user need to click open a link to access photos for comparison. Our work builds on user trust, we’ll do our best to make sure our users have trust in what we do. Plus, this improves usability by giving user preview to returned photos :)

Talked to a mother who is still searching for her son lost more than 20 years ago

Painfully admitting there is little we could help her at this moment. But our belief will still bring us forward.

Good Design Award Second Screening

We entered second round of screening for this year's Good Design Award! Our sincere appreciation to this year's juries and our supporters!

Weibo verified our App :)

After at least a dozen attempts, Sina Weibo (新浪微博) finally verified our App. This will give us better access to Weibo's APIs, and more impressions to our posts, hopefully.

We are in SAP's 1 billion lives project

We submitted Zhongxun to SAP's 1 billion lives project initiative, and it has been voted as one of ten candidates :). The presentation will be on 21.07.2017, come to support us!

zhongxun.life was blocked by our web hosting company in China :(

This is due to ICP regulation, a bit crazy, but we will solve it. For the moment, zhongxun.life will be redirected to another temporary server, and we have to use IP address in auto-generated links :(

Talked to Baobeihuijia (宝贝回家) volunteers

We have been receiving photos from Baobeihuijia (宝贝回家) volunteers via Weibo chat for face recognition related tasks. Great admiration for their work. We want to create the best tool for them!

zhongxun.life's admin portal is online!

We have observed substantial noise in user submitted photos. The admin portal will give us better control over photo quality.

closing the learning loop!

Zhongxun can now automatically post photos it thinks similar to Weibo, and incorporate user feedback into its learning loop. Yeeha :)

Zhongxun is now integrated with Weibo (新浪微博)!

You can now send photo for face recognition to Zhongxun via chat function in Weibo! Find Zhongxun @ '人脸识别寻找被拐儿童' on Weibo :)

Stopped that nasty DDOS Attack

Our server was under DDOS attack in the past week. We fixed that.

We received our 1st fans on Weibo!

absolutely thrilled :)

Launched Zhongxun's first test version on Weixin (微信) sandbox

Help us test Zhongxun's face recognition function by sending photos to its Weixin sandbox. Thank you!!

Contact Us