SchGAN:小学校园布局生成,建筑设计智能助手

- AI小学校园布局生成助手

SchGAN(测试版)

小学校园的规划布局是方案设计前期确定深化方向的重要环节。一个合理的小学校园布局设计,既要考虑校园主要出入口的位置,教学楼、运动场、体育馆等功能布局与相互位置关系的合理性,同时还要满足教学楼的噪声间距、楼层限高、朝向等多方面要求,这通常需要依赖设计师的经验,并进行反复的测算与比较,以寻求一个能满足多种需求的最优布局。

The layout of the primary school campus is an important part of the preliminary design to determine the deepening direction. A reasonable layout design of the primary school campus should not only consider the location of the main entrances and exits of the campus, but also the rationality of the functional layout and the relationship between each other in the teaching building, sports ground, gymnasium and other functional layouts. At the same time, it also needs to meet the noise spacing, floor height, orientation and other requirements, which usually depends on the designers experience, and repeatedly measured and compared in order to find an optimal layout to meet a variety of needs.

SchGAN是一个运用真实小学案例样本完成对抗生成训练的智能辅助设计系统。根据用户在INPUT方框内自由绘制的周边道路(道路等级、道路形状)以及用地范围(矩形、多边形),经过SchGAN运算处理,在OUTPUT方框中,数秒钟之内就可以帮助设计师自动生成基于该场地条件的小学校园规划布局方案。(注:所有生成的结果都默认为上北下南)

SchGAN is an intelligent aided design system which uses real primary school case samples to complete antagonistic generation training. According to the surrounding road (road grade, road shape) and the land range (rectangle, polygon) freely drawn by the user in the INPUT box. After SchGAN operation processing, in the OUTPUT box, a few seconds can help designers automatically generate based on the site conditions of the primary school campus planning layout scheme. (Note: All generated results default to up as north )

SchGAN生成结果

SchGAN的训练过程完全是基于对真实小学样本中设计经验的自我统计学习,没有任何建筑设计原理或规范等规则的输入。在对测试样本进行模型效果验证时,我们发现SchGAN生成的设计方案甚至优于部分真实方案的设计,这让我们看到了机器学习在建筑生成设计中的无限潜力和可能性。

SchGANs training process is entirely based on self-statistical learning of design experience in real primary school samples, without any input from rules such as architectural design principles or codes. We find that SchGANs design is even better than that of some real ones, which shows us the infinite potential and possibilities of machine learning in architectural design.

SchGAN动画操作演示过程

SchGAN网页交互界面和用户体验流程详细输入步骤

- SchGAN详细输入步骤

1.在Input方框内绘制周边道路及用地范围:


(1)选择“直线”或“曲线”,即选择道路形状;
(2)选择“主路”、“次路”或者“支路”,即选择道路等级与宽窄;
(3)在方框内绘制道路

注意:绘制直线道路时,按住鼠标左键不放开,拖动到特定位置后松开;绘制曲线道路时,先确定道路头尾两点,再确定中间点,即先按住鼠标左键不放开,拖动到指定位置后松开,再在方框内任一空白处点击鼠标左边然后拖动道路,调整到合适形状后松开


(4)在道路基础上绘制用地

注意:绘制矩形时,鼠标左键按住不放开,拖动到合适位置后松开;绘制多边形时,鼠标左键点击第一个点后松开,拖到其他空白处再点击左键绘制第二个点后松开,以此类推绘制接下来的控制点,当绘制完成时,将鼠标箭头移动至方框外即完成绘制。

2.点击“智能生成”按键,上传已绘制好的基地条件


3.在Output方框内查看SchGAN完成的设计结果


4.对SchGAN的设计结果进行评价

注意:先点击“专业人士”或“非专业人士”,再按五分制评分

SchGAN detailed input steps:

1. Draw the perimeter road and land use in the Input box;

(1) Select straight line or curve, i.e. select the road shape;

(2) Choosing main road, secondary road or branch road, i.e. choosing road grade and width;

(3) Draw the road in the box

Note: When drawing a rectangle, the left mouse button should be  held down until it is dragged to the appropriate position; When drawing a curved road, first determine the head and end of the road two points, and then determine the middle point, that is,  press the left mouse button from the starting point to the end point first, and then press any blank in the box to drag the road to adjust to the appropriate shape;

(4) Draw the land on the basis of the road 

Note: the left mouse button should be held down until it is dragged to the appropriate position; When drawing a polygon, click the first control point, then click the second control point, and so on to draw the next control points. When the drawing is complete, move the mouse arrow out of the box to end the drawing.

2. Click the ‘intelligent generation’ button to upload the plotted base conditions

3. View the design results SchGAN completed in the Output box

4. Evaluate SchGAN’s design results 

Note: click on Professional or Non-professional and score on a 5-point scale



- SchGAN测试版试用链接

http://lll702.wicp.vip

建议使用火狐浏览器、360高速浏览器

SchGAN Beta Trial Link: http://lll702.wicp.vip/ 

Firefox and 360 High Speed Browser are recommended

作者简介:

刘宇波,华南理工大学建筑学院建筑系主任、教授、博导,哈佛大学公派访问学者,liuyubo@scut.edu.cn

邓巧明,华南理工大学建筑学院副教授、硕导,美国麻省理工学院MIT访问学者,dengqm@scut.edu.cn

梁凌宇,华南理工大学电信学院副教授、硕导,lianglysky@gmail.com

联系作者也可以关注公众号(设计未来实验室)

同时,欢迎在mixlab-无界社区一起探索AI+

评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值