News classification
Contact us
- Add: No. 9, North Fourth Ring Road, Haidian District, Beijing. It mainly includes face recognition, living detection, ID card recognition, bank card recognition, business card recognition, license plate recognition, OCR recognition, and intelligent recognition technology.
- Tel: 13146317170 廖经理
- Fax:
- Email: 398017534@qq.com
Depth learning into development artifact
Depth learning into development artifact
Recently, Uizard Technologies, a start-up company based in Copenhagen, has developed a system called pix2code. With depth learning, the system can directly output the corresponding code based on the input graphical user interface (GUI) screenshots, eliminating the need for manual writing of code at the front.
At present, the accuracy of pix2code generated code has reached 77%, and these codes can be compatible with Android, iOS and Web interface three platforms.
How amazing is it? Take a look at the video below.
"In order to produce this neural network, the research team has to overcome three major difficulties," says Tony Beltramelli, founder of UIzard:
The first is the computer vision level the computer can't automatically recognize and understand the given objects and scenes, nor can they determine their location and features;
The second is the language level - they need to teach the neural network to understand the text so that it can create accurate samples;
In addition, on GitHub, Beltramelli answers some common questions related to pix2code. So part of the compiled as follows:
Q: when will the data meeting be open?
A: we put the paper in this year's NIPS, and when they give or reject the results, the data set will be open, the specific time in September. We'll then provide GUI screenshots, related DSL code, and three target codes for the iOS and An Zhuohe Web interfaces that are included in the data set.
Q: when is the source code open?
A: originally, as written in the paper, we have no plans to open the source code. But I didn't expect this project to attract so much attention, so we decided to open the pix2code implementation code and data set together in the paper.
Does Q:pix2code support other target platforms / languages?
A: No, pix2code is just a research project that will keep the state described in the paper. This project is just a small display of what we've done at Uizard Technologies. Of course, we welcome you, fork, to experiment on other platforms / languages on your own.
Q: can I use pix2code on my front end project?
A: No, pix2code is just an experimental project, and it's not going to allow you to apply it to a particular case at the moment. But we are working hard to make it commercial.
Q: how is the performance of the model measured?
A: the accurate or wrong results reported in the paper are all on the DSL level by comparing the generated token with the expected token. If there is any difference in length between the two, it is also considered to be wrong.
Q: how long does it take to train the model?
A: in a piece of NVIDIA's Tesla K80 GPU, to make a data set including 109 * 10^6 parameter optimization, it takes less than 5 hours. So if you want to train the model on three target platforms, it will take about 15 hours.
Q: I'm a front end developer Am I going to lose my job soon? I asked this question very sincerely, a lot of times
A:AI won't replace the front-end Engineer so soon.
If there is already a mature version of pix2code, in each of the different platform / language generated code can achieve 100% accuracy, good front still need logic, interactive, advanced graphics and animation, and all other users of love.
We do this to fill the gap between UI/UX designers and front-end developers rather than replace them. We want designers to create better, and allow developers to spend more of their time on those core functions.
We believe that in the future, AI will work with humanity instead of human beings.
At present, the accuracy of pix2code generated code has reached 77%, and these codes can be compatible with Android, iOS and Web interface three platforms.
How amazing is it? Take a look at the video below.
"In order to produce this neural network, the research team has to overcome three major difficulties," says Tony Beltramelli, founder of UIzard:
The first is the computer vision level the computer can't automatically recognize and understand the given objects and scenes, nor can they determine their location and features;
The second is the language level - they need to teach the neural network to understand the text so that it can create accurate samples;
Finally, the neural network needs to understand the code, the text, and the correlation between the corresponding images.
In addition, on GitHub, Beltramelli answers some common questions related to pix2code. So part of the compiled as follows:
Q: when will the data meeting be open?
A: we put the paper in this year's NIPS, and when they give or reject the results, the data set will be open, the specific time in September. We'll then provide GUI screenshots, related DSL code, and three target codes for the iOS and An Zhuohe Web interfaces that are included in the data set.
Q: when is the source code open?
A: originally, as written in the paper, we have no plans to open the source code. But I didn't expect this project to attract so much attention, so we decided to open the pix2code implementation code and data set together in the paper.
Does Q:pix2code support other target platforms / languages?
A: No, pix2code is just a research project that will keep the state described in the paper. This project is just a small display of what we've done at Uizard Technologies. Of course, we welcome you, fork, to experiment on other platforms / languages on your own.
Q: can I use pix2code on my front end project?
A: No, pix2code is just an experimental project, and it's not going to allow you to apply it to a particular case at the moment. But we are working hard to make it commercial.
Q: how is the performance of the model measured?
A: the accurate or wrong results reported in the paper are all on the DSL level by comparing the generated token with the expected token. If there is any difference in length between the two, it is also considered to be wrong.
Q: how long does it take to train the model?
A: in a piece of NVIDIA's Tesla K80 GPU, to make a data set including 109 * 10^6 parameter optimization, it takes less than 5 hours. So if you want to train the model on three target platforms, it will take about 15 hours.
Q: I'm a front end developer Am I going to lose my job soon? I asked this question very sincerely, a lot of times
A:AI won't replace the front-end Engineer so soon.
If there is already a mature version of pix2code, in each of the different platform / language generated code can achieve 100% accuracy, good front still need logic, interactive, advanced graphics and animation, and all other users of love.
We do this to fill the gap between UI/UX designers and front-end developers rather than replace them. We want designers to create better, and allow developers to spend more of their time on those core functions.
We believe that in the future, AI will work with humanity instead of human beings.