Competition is introduced

When industrial pollution is rampant, have you ever thought of using AI to identify the sources of pollution scattered all over the place?

When it has become common knowledge to pay attention to left-behind children and the elderly, have you ever thought of using AI to ensure the safety of your family?

As doctors struggle to spot patients’ underlying symptoms, have you ever thought of using AI to identify hidden symptoms?

Too big a programming task? Don’t be afraid! EasyDL is a customized model training platform launched by Baidu. It does not need algorithm programming, but can train customized machine learning model through simple page operation and a small amount of data. As long as you are creative, come and join us!

The registration procedure

1. Fill in the intention of registration

Click on me to fill out the application form

2. Join the wechat group of the contest

The latest information of this contest will be published in the group in real time, please be sure to enter the group

3. Train the model through EasyDL platform and release it

EasyDL Platform Login Address (computer only)

EasyDL model training tutorial

4. Submit the model

Fill in the submission form after the model has been successfully published

Click Fill in Form to submit the model

Competition schedule

  • Submission of models for the preliminary contest began on October 10
  • Nov. 10 deadline for submission of preliminary contest models
  • November 14, the rematch list announced
  • 17 November Final (Semi-final)

Competition system:

Preliminary competition time: October 10th to November 10th

Technology has become an indispensable source of change in the world. Brain industry application innovation challenge on baidu baidu EasyDL customized training platform of image recognition, voice recognition function point, and aims to promote technology innovation, geared to the needs of individuals, groups and enterprise developers invitation, use a small amount of data and quick training integration platform, widely problems the industry China, fu to the world, to the future.

The final list is announced: November 14

The final match will be held in Beijing, Suzhou and Shenzhen on November 17th.

Contestants are divided according to the competition area, and should bring a complete industry landing plan (including industry problem elaboration, model content, executable landing plan, etc.) to explain at the final site. The on-site jury will give a comprehensive score based on commercial value, feasibility, model conception, data difficulty and other dimensions, and then decide the ranking and award the prize.

Division of competition:

East China, central China, south China

Final City:

Beijing, Suzhou, Shenzhen

Competition area and corresponding final city

The north China

Beijing, Tianjin, Hebei, Shanxi, Inner Mongolia, Liaoning, Jilin, Heilongjiang, Ningxia, Xinjiang, Qinghai, Shaanxi, Gansu

Central China

Hubei, Hunan, Henan, Jiangxi, Shandong, Jiangsu, Anhui, Shanghai, Zhejiang, Fujian

South China

Sichuan, Yunnan, Guizhou, Xizang, Chongqing, Guangdong, Guangxi, Hainan, Hong Kong, Macao, Taiwan

Tournament bonus

North China, Central China, south China three regions on the same day of the run-off, each region respectively decided

1 first prize, 20,000 yuan cash bonus +6.6W Baidu cloud storage virtual resources.

2 second prize, 15,000 yuan cash prize.

Third prize 3, 8 thousand yuan cash prize.

Case sharing

1. Retail scenario

Artificial intelligence technology has gradually made its way into the retail space, but there are still a lot of problems in the segmentation scene that need to be solved. Please train the scheme model for specific problems according to the actual requirements in the production and operation process.

Case indication: shelf inspection

By connecting with EasyDL customized image open platform technology, Hangzhou Huihe Company helps E-Dianjia platform to implement commodity display audit and upgrade of traditional retail stores, reconstructs FMCG industry’s participation in store implementation activities, and promotes the digital transformation of traditional FMCG brands and national retail stores.

2. Medical scenario

The reform of medical scenarios in the application of existing scientific and technological innovations in the industry has attracted much attention. Please combine the specific problems of medical scenarios such as imaging diagnosis, document verification, data analysis and so on to conduct scheme model training.

Case study: X-ray chest disease recognition

Guangzhou kaihui information through the training of lung imaging images, to obtain the model that can identify lung diseases. This model can identify the shadow features of lung X-ray, assist doctors to judge pneumonia and lung cancer, and reduce the occurrence of misdiagnosis and missed diagnosis.

3. Security scenario

The topic of “safety” exists in every corner of the real production and life, from home security to public area security, from the protection of children to the safety of the elderly, or from production safety to road detection, please carry out program model training according to the actual needs.

Case reference: Check external circuit hazards

The Foshan Power Supply Bureau of China Southern Power Grid trained the pictures of cranes, excavators and other situations near the transmission line, and obtained the corresponding model after multiple optimization. This model can identify the external hidden dangers of transmission lines, and automatically report the problems found, so as to ensure the safe operation of transmission lines.

4. Industrial quality inspection

The high labor cost of quality inspection industry has always been an important topic discussed in the industry. How to bring the highly efficient ARTIFICIAL intelligence technology into the traditional industry and save resources for actual production, please combine the industry insight to conduct model training.

Case reference: keyboard assembly qualification test

Dieyu Technology in the manufacturing and assembly of keyboard assembly line, identify the qualification of the keyboard assembly, including missing parts, wrong, normal three categories, baidu EasyDL interface and industrial camera, industrial light source, laser rangefinder, PLC control cylinder and other equipment to form a comprehensive detection system.

5. Select a scenario

Contestants can create and train models for other industry scenarios. The models must be combined with real pain points and actual problems in production and life, and specific explanations should be made in the registration information.

For more information

Want to learn more about the tournament? Welcome to group communication.