Codelabs & Seminar, small corner, great harvest!

A little, a great gain!

Come with me to see two codelabs in conference room 5 of the Science and Art Center, and the teachers’ seminar

At 13:20, the “Build Your First Android App in Kotlin” Codelab was delivered by Engineer Wu Dewen of Xiamen University. Not only students from the School of Information, but also students from other colleges and universities were present.

This is an experiment on Google’s Codelab website. Through this experiment, Wu points out many differences between Kotlin and Java.

First of all, the present students began to follow the teacher’s explanation and set up an Android App project. Since most of the participants were developing Android App for the first time, the teacher explained each step of the new project in detail.

Download androidstudio, plugins, and more from dl.google.com.

Then, Teacher Wu Dewen explained in detail the use of Layout editor and Android code in the project development process, focusing on the syntax and advantages of Kotlin.

Tips: Lesson plan View: Search Codelab China

In plain English, Kotlin is trying to get developers to type as few letters as possible, Wu said. For example, “function” is shortened to “fun”; Code auto-completes by hitting enter, eliminating the semicolon at the end of each line and “taking at least two seconds less”, greatly increasing efficiency. In addition, Kotlin’s website (kotlinlang.org) says, “Don’t block Keep moving,” reflecting its focus on efficiency.

“Finally, welcome to be a happy code farmer.”

Due to time constraints, the teacher will be some simple or not too critical content, so remind students to carefully read the content of the lesson plan, will deal with the details of the problem. Rao is so, more than an hour of lecture, still make the students benefit a lot, more.

At 14:40, participants with computers equipped with The Codelab environment looked on intently and started sharing.

The audience is full of dry goods:

“There are two main methods of downsampling.”

“This function-like approach is called functional API in Keras.”

“You can handle 256 or 512.”

“The purpose of passing in a random SEED is to make every run the same.”

…

Through the training process demonstrated by the lecturer Li Zhuoru, the participants found that the model gradually fitted, and eventually dropped to 0.0043.

We will test whether the model can be imported successfully, run the code, put what we have learned into practice in time, and experience the process of extracting the lines in the color cartoon pictures and generating the black and white line draft.

During the sharing period, students communicated enthusiastically for many times and digested what they learned.

Dr Li also notes that it is possible to colour black and white images with what he has learned using a neural network, but that would require much more data.

Preprocessing lines of different thickness can reduce the amount of machine learning. On the contrary, line stylization model processing can make lines have changes in thickness, more aesthetic. The sharer showed a set of manipulated image comparisons to make the model’s effects and advantages more obvious. Next, he shows the training process, saving the model and using the model.

The atmosphere was very harmonious, and there were questions from time to time. Dr. Li Zhuoru explained patiently, making the sharing more like a communication conference.

The meeting ended with laughter when Dr. Li shared a cartoon drawn by a drawing robot.

After the meeting, the students gathered around the sharer, and actively asked questions and communicated.