OpenCV Seminar (May 18, 2019)

The OpenCV seminar was successfully held in Shenzhen on May 18th, and 80 people attended the conference. The speakers of the seminar were:

• Gary Bradski, the founder of OpenCV and author of "Learning OpenCV";
• Vadim Pisarevsky, the head of OpenCV development team;
• Shiqi Yu, the head of OpenCV Chinese community;
• Zhou Zhaojing, Intel senior Application Engineer;
• Dmitry Kurtaev, member of the OpenCV Development Team Deep Learning Module.

Gary Bradski gave a report entitled "Taking the Red Pill: AI, Robotics and the Nature of the Mind" with the allusion of the red pills and blue pills in the Matrix movie, discussing artificial intelligence, robotics and human perception.

Gary Bradski in presentation

Discussion after report

Vadim Pisarevsky has been leading the development of OpenCV since the beginning of OpenCV. His report first introduced the new features of OpenCV, and then presented the code optimization method, which gradually speeded up the example of a convolution operation by 300 times.

Vadim Pisarevsky in presentation

Shiqi Yu, the designer of the fast face detection algorithm, introduced the design idea of the open source libfacedetection library. Then he shared a series of algorithm optimization methods such as model simplification, SIMD optimization, code optimization, fixed point and so on. In addition, he also introduced the libfacedetection model to achieve 4 times faster speed under OpenVINO acceleration.

Yu Shiqi in presentation

In the afternoon, Intel China engineer Zhaojing Zhou introduced the OpenVINO tool, which can optimize the trained models and convert them into IR file format. In the OpenVINO tool, optimized code for the mainstream network layer and mainstream models is built in. OpenVINO's superior performance, deep learning deployment can be completely deployed on the OpenVINO+Intel hardware platform, without independent GPU. At the same time, OpenCV is also compatible with OpenVINO. Meanwhile, Zhaojing also guided everyone to use DevCloud through the browser to use the various hardware platforms in the cloud to run deep learning algorithms.

Zhou Zhaojing in presentation

Dmitry Kurtaev, who is responsible for deep learning module development in the OpenCV team, introduced OpenCV's deep learning module, OpenVINO's acceleration of OpenCV, presenting the application in the form of sample code, and showing how it can achieve higher performance.

Dmitry Kurtaev is reporting

The seminar was held from 9:00 am to 6:00 pm in a lively atmosphere! The participants and speakers of the seminar had a heated discussion together.

The reporter PPT will be uploaded to http://www.opencv.org.cn/px/

Later OpenCV China team will launch more activities, please pay attention to the following public number and get notice.