Cumincad : CUMINCAD Papers : Search Results

CumInCAD is a Cumulative Index about publications in Computer Aided Architectural Design
supported by the sibling associations ACADIA, CAADRIA, eCAADe, SIGraDi, ASCAAD and CAAD futures

PDF papers
References

Hits 1 to 20 of 653

Reformat results as: short short into frame detailed detailed into frame

_id	ecaade2020_290
id	ecaade2020_290
authors	Elesawy, Amr Alaaeldin, Signer, Mario, Seshadri, Bharath and Schlueter, Arno
year	2020
title	Aerial Photogrammetry in Remote Locations - A workflow for using 3D point cloud data in building energy modeling
source	Werner, L and Koering, D (eds.), Anthropologic: Architecture and Fabrication in the cognitive age - Proceedings of the 38th eCAADe Conference - Volume 1, TU Berlin, Berlin, Germany, 16-18 September 2020, pp. 723-732
doi	https://doi.org/10.52842/conf.ecaade.2020.1.723
summary	Building energy modelling (BEM) results are highly affected by the surrounding environment, due to the impact of solar radiation on the site. Hence, modelling the context is a crucial step in the design process. This is challenging when access to the geometrical data of the built and natural environment is unavailable as in remote villages. The acquisition of accurate data through conventional surveying proves to be costly and time consuming, especially in areas with a steep and complex terrain. Photogrammetry using drone-captured aerial images has emerged as an innovative solution to facilitate surveying and modeling. Nevertheless, the workflow of translating the photogrammetry output from data points to surfaces readable by BEM tools proves to be tedious and unclear. This paper presents a streamlined and reproducible approach for constructing accurate building models from photogrammetric data points to use for architectural design and energy analysis in early design stage projects.
keywords	Building Energy Modeling; Photogrammetry; 3D Point Clouds; Low-energy architecture; Multidisciplinary design; Education
series	eCAADe
email
last changed	2022/06/07 07:55

_id	ecaade2020_120
id	ecaade2020_120
authors	Ishikawa, Daichi, Fukuda, Tomohiro and Yabuki, Nobuyoshi
year	2020
title	A Mixed Reality Coordinate System for Multiple HMD Users Manipulating Real-time Point Cloud Objects - Towards virtual and interactive 3D synchronous sharing of physical objects in teleconference during design study
source	Werner, L and Koering, D (eds.), Anthropologic: Architecture and Fabrication in the cognitive age - Proceedings of the 38th eCAADe Conference - Volume 1, TU Berlin, Berlin, Germany, 16-18 September 2020, pp. 197-206
doi	https://doi.org/10.52842/conf.ecaade.2020.1.197
summary	Teleconferences without travel costs are useful for building a consensus in design. However, conventional teleconferencing using computer displays and web cameras is well known to have performance problems due to the lack of co-presence feeling with remote participants and the difficulty in sharing three dimensional (3D) information intuitively. This research proposes a method to share the mixed reality (MR) coordinate system for multiple head-mounted display (HMD) users manipulating real-time point cloud objects for the virtual and interactive 3D synchronous sharing in teleconferences. In our proposed method, the reference point of the virtual world coordinate system called world anchor and local coordinates of segmented point cloud objects in real-time are shared among HMDs via a server PC to share the same MR coordinate system. Using this method, the result of moving and rotating manipulation using hand gestures for segmented point cloud objects by an HMD user are reflected in the other HMD users. We developed a prototype system and evaluated the performance of the system when multiple users used this system. Future works include adapting this system to multiple RGB-D cameras and the internet environment.
keywords	Mixed reality coordinate system; Real-time point clouds; Multiple User Interaction; Teleconference; 3D Synchronous Physical Object Sharing
series	eCAADe
email
last changed	2022/06/07 07:50

_id	sigradi2020_60
id	sigradi2020_60
authors	Asmar, Karen El; Sareen, Harpreet
year	2020
title	Machinic Interpolations: A GAN Pipeline for Integrating Lateral Thinking in Computational Tools of Architecture
source	SIGraDi 2020 [Proceedings of the 24th Conference of the Iberoamerican Society of Digital Graphics - ISSN: 2318-6968] Online Conference 18 - 20 November 2020, pp. 60-66
summary	In this paper, we discuss a new tool pipeline that aims to re-integrate lateral thinking strategies in computational tools of architecture. We present a 4-step AI-driven pipeline, based on Generative Adversarial Networks (GANs), that draws from the ability to access the latent space of a machine and use this space as a digital design environment. We demonstrate examples of navigating in this space using vector arithmetic and interpolations as a method to generate a series of images that are then translated to 3D voxel structures. Through a gallery of forms, we show how this series of techniques could result in unexpected spaces and outputs beyond what could be produced by human capability alone.
keywords	Latent space, GANs, Lateral thinking, Computational tools, Artificial intelligence
series	SIGraDi
email
last changed	2021/07/16 11:48

_id	caadria2020_234
id	caadria2020_234
authors	Zhang, Hang and Blasetti, Ezio
year	2020
title	3D Architectural Form Style Transfer through Machine Learning
source	D. Holzer, W. Nakapan, A. Globa, I. Koh (eds.), RE: Anthropocene, Design in the Age of Humans - Proceedings of the 25th CAADRIA Conference - Volume 2, Chulalongkorn University, Bangkok, Thailand, 5-6 August 2020, pp. 659-668
doi	https://doi.org/10.52842/conf.caadria.2020.2.659
summary	In recent years, a tremendous amount of progress is being made in the field of machine learning, but it is still very hard to directly apply 3D Machine Learning on the architectural design due to the practical constraints on model resolution and training time. Based on the past several years' development of GAN (Generative Adversarial Network), also the method of spatial sequence rules, the authors mainly introduces 3D architectural form style transfer on 2 levels of scale (overall and detailed) through multiple methods of machine learning algorithms which are trained with 2 types of 2D training data set (serial stack and multi-view) at a relatively decent resolution. By exploring how styles interact and influence the original content in neural networks on the 2D level, it is possible for designers to manually control the expected output of 2D images, result in creating the new style 3D architectural model with a clear designing approach.
keywords	3D; Form Finding; Style Transfer; Machine Learning; Architectural Design
series	CAADRIA
email
last changed	2022/06/07 07:57

_id	ijac202018103
id	ijac202018103
authors	Kimm, Geoff
year	2020
title	Actual and experiential shadow origin tagging: A 2.5D algorithm for efficient precinct-scale modelling
source	International Journal of Architectural Computing vol. 18 - no. 1, 41-52
summary	This article describes a novel algorithm for built environment 2.5D digital model shadow generation that allows identities of shadowing sources to be efficiently precalculated. For any point on the ground, all sources of shadowing can be identified and are classified as actual or experiential obstructions to sunlight. The article justifies a 2.5D raster approach in the context of modelling of architectural and urban environments that has in recent times shifted from 2D to 3D, and describes in detail the algorithm which builds on precedents for 2.5D raster calculation of shadows. The algorithm is efficient and is applicable at even precinct scale in low-end computing environments. The simplicity of this new technique, and its independence of GPU coding, facilitates its easy use in research, prototyping and civic engagement contexts. Two research software applications are presented with technical details to demonstrate the algorithm’s use for participatory built environment simulation and generative modelling applications. The algorithm and its shadow origin tagging can be applied to many digital workflows in architectural and urban design, including those using big data, artificial intelligence or community participative processes.
keywords	2.5D raster, actual and experiential shadow origins, generative techniques, participatory built environment simulation, reactive scripting for design
series	journal
email
last changed	2020/11/02 13:34

_id	caadria2020_062
id	caadria2020_062
authors	Lu, Ming and Yuan, Philip F.
year	2020
title	A New Algorithm to Get Optimized Target Plane on 6-Axis Robot For Fabrication
source	D. Holzer, W. Nakapan, A. Globa, I. Koh (eds.), RE: Anthropocene, Design in the Age of Humans - Proceedings of the 25th CAADRIA Conference - Volume 2, Chulalongkorn University, Bangkok, Thailand, 5-6 August 2020, pp. 393-402
doi	https://doi.org/10.52842/conf.caadria.2020.2.393
summary	In usual robotic fabrication by 6 axis industrial robot such as KUKA ,ABB and other brands ,the usual robot's 4th ,5th and 6th axis is exactly converge in one point .When this type robot (pieper) is doing movement commands ,setting the degree of 4th axis close to zero is an ideal condition for motion stability ,especially for putting device which connect to tool head on 4th axis arm part.In plastic melting or others print which not cares the rotation angle about the printing direction(the printing direction means the effector's output normal direction vector, KUKA is X axis,ABB is Z axis) ,the optimization of 4th axis technology not only makes printing stable but also makes better quality for printing.The paper introduces a new algorithm to get the analytics solution.The algorithm is clear explained by mathematics and geometry ways. At the end of paper, a grasshopper custom plugin is provided ,which contains this new algorithm ,with this plugin, people can get the optimized target path plane more easily.
keywords	3D printing; brick fabrication; robotic; optimization algorithm; grasshopper plugin
series	CAADRIA
email
last changed	2022/06/07 07:59

_id	ecaade2020_264
id	ecaade2020_264
authors	Nicholas, Paul, Rossi, Gabriella, Papadopoulou, Iliana, Tamke, Martin, Aalund Brandt, Nikolaj and Jessen Hansen, Leif
year	2020
title	Precision Partner - Enhancing GFRC craftsmanship with industry 4.0 factory-floor feedback
source	Werner, L and Koering, D (eds.), Anthropologic: Architecture and Fabrication in the cognitive age - Proceedings of the 38th eCAADe Conference - Volume 2, TU Berlin, Berlin, Germany, 16-18 September 2020, pp. 631-640
doi	https://doi.org/10.52842/conf.ecaade.2020.2.631
summary	This paper presents a novel human-machine collaborative approach to automatic quality-control of Glass-Fiber Reinforced Concrete (GFRC) molds directly on the factory floor. The framework introduces Industry 4.0 technologies to enhance the ability of skilled craftsmen to make molds through the provision of horizontal feedback regarding dimensional tolerances. Where digital tools are seldom used in the fabrication of GFRC molds, and expert craftsmen are not digital experts, our implementation of automated registration and feedback processes enables craftsmen to be integrated into and gain value from the digital production chain. In this paper, we describe the in-progress framework, Precision Partner, which connects 3d scanning and point cloud registration of geometrically complex and varied one off elements to factory floor dimensional feedback. We firstly introduce the production context of GFRC molds, as well as industry standards for production feedback. We then detail our methods, and report the results of a case study that tests the framework on the case of a balcony element.
keywords	3d Scanning; GFRC; Feedback; Automation; Human in the loop; Digital Chain
series	eCAADe
email
last changed	2022/06/07 07:58

_id	acadia20_160p
id	acadia20_160p
authors	Scelsa, Jonathan A.; Birkeland, Jennifer
year	2020
title	The Collective Perspective Machine
source	ACADIA 2020: Distributed Proximities / Volume II: Projects [Proceedings of the 40th Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-95253-6]. Online and Global. 24-30 October 2020. edited by M. Yablonina, A. Marcus, S. Doyle, M. del Campo, V. Ago, B. Slocum. 160-163
summary	Since the age of humanism, both on the easel and our screens, the production of the architectural image has been conventionally governed by one individual, whom we might refer to as the drafter. As the primary author sitting in the chair of the vantage point, the drafter occupies the privileged position, for whom the translation between the second and third dimensions establishes an approximate realism. The viewers, or secondary participants, by contrast, are relegated to a subordinate position, subject to the residual distortions of the drafter’s vision, based on their relative vantage points. While perhaps cynical, our current condition does not share the same philosophical positivistic optimism of the Renaissance, nor the ideal faith in humanity that empowered the democratic universalisms of modernity. Rather, it is formed from an ambiguous inquiry into creating a new sense of truth, brought forth by the proliferation and amplification of multiple individual ‘perspectives.’ In his conclusion to The Projective Cast, Evans illustrates ten ‘transitive spaces’ of geometric projection towards the generation and representation of a designed object. The fifth “transitive space” describes the space between a building or object and its defined perspectival representations. Evans observes that this path typically follows the progression from the object to a photo or a drawing and is rarely reversed. This project and machine designed for an exhibition seeks to establish a new procedure for generating design, neither subjectively from a personal static individual point nor objectively in the round for all to experience equally. Instead, a new machine establishes form as the hybrid of multiple responsive perspectives wherein all viewers are simultaneously the generator of projective form and the receiver of distorted images.
series	ACADIA
type	project
email
last changed	2021/10/26 08:03

_id	acadia20_198
id	acadia20_198
authors	Sinke Baranovskaya, Yuliya; Tamke, Martin; Ramsgaard Thomsen, Mette
year	2020
title	Simulation and Calibration of Graded Knitted Membranes
source	ACADIA 2020: Distributed Proximities / Volume I: Technical Papers [Proceedings of the 40th Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-95213-0]. Online and Global. 24-30 October 2020. edited by B. Slocum, V. Ago, S. Doyle, A. Marcus, M. Yablonina, and M. del Campo. 198-207.
doi	https://doi.org/10.52842/conf.acadia.2020.2.198
summary	The grading of knit changes its geometrical performance and steers membrane expansion. However, knit possesses challenges of material predictability and digital simulation, due to its multiscalar complexity and anisotropic properties. Taking as a challenge the lack of digital solutions incorporating CNC-knit performance into the design model, this paper presents a novel approach for the design-integrated simulation of graded knit, informed by an empirical dataset analysis in combination with genetic optimization algorithms. Here the simulation design tool reflects the differences of industrially knitted textile panel behavior through digital mesh grading. Diversified fabric stiffness is achieved by intertwining the yarn into variegated stitch types that steer the textile expansion under load. These are represented digitally as zoned quad meshes with each segment assigned a stiffness value. Mesh stiffness values are optimized by minimizing the distance between the point clouds and the digital mesh, which are documented through deviation colored maps. This work concludes that design properties—pattern topology, stitch ratio, pattern density—play an important role in textile panel performance under load. Stiffness values derived from the optimization are higher for shallower designs and lower for the deeper cones.
series	ACADIA
type	paper
email
last changed	2023/10/22 12:06

_id	sigradi2020_563
id	sigradi2020_563
authors	Souza, Thiago Leitao de; Filho, Giovany Bicalho de Lourdes; Silva, Gustavo Lennon da Silva; Silva, Vinicius da Conceiçao
year	2020
title	A 360° history of the city: the digital reconstruction of the Rio de Janeiro Panorama by Victor Meirelles and Henri Langerock from the end of the 19th century
source	SIGraDi 2020 [Proceedings of the 24th Conference of the Iberoamerican Society of Digital Graphics - ISSN: 2318-6968] Online Conference 18 - 20 November 2020, pp. 563-568
summary	This work is an investigation of “O Panorama do Rio de Janeiro, by Victor Meirelles and Henri Langerock”, aiming at its digital reconstitution for 360? immersion activities. To achieve this goal, digital and analog systems of representations will be developed and applied, including: computer graphics techniques, pictorial layers, 3D models, 3D renderings and freehand drawings. Emphasizing that through practical and theoretical investigations, the work developed is a new 360? interpretation of the artists' painting.
keywords	Panorama of Rio de Janeiro, 360 ° Immersive experience, City history, Virtual reality
series	SIGraDi
email
last changed	2021/07/16 11:52

_id	ecaade2020_108
id	ecaade2020_108
authors	Steino, Nicolai
year	2020
title	Post-Conflict Reconstruction - Small scale elements of a parametric urban design approach
source	Werner, L and Koering, D (eds.), Anthropologic: Architecture and Fabrication in the cognitive age - Proceedings of the 38th eCAADe Conference - Volume 2, TU Berlin, Berlin, Germany, 16-18 September 2020, pp. 69-78
doi	https://doi.org/10.52842/conf.ecaade.2020.2.069
summary	Taking the Syrian city of Homs as its point of departure, this paper aims to suggest some first components of a parametric urban design approach to post-conflict reconstruction focused on scenario building. From analyses of social, physical and environmental infrastructures and theoretical positions on environmentally and socially sustainable urban design in the Middle Eastern culture and climate, a framework and some initial tests for a parametric 3D urban model developed in CityEngine are presented. With the intended purpose of providing a tool capable of visualising modifiable urban design scenarios along with relevant associated data, the presented work focuses on the smallest scale of a model encompassing scales from the district level to the level of the urban block, with some relevant architectural features relating to social and environmental qualities.
keywords	parametric urban design; post-conflict reconstruction; scenario building; climate-adaptive design
series	eCAADe
email
last changed	2022/06/07 07:56

_id	caadria2020_154
id	caadria2020_154
authors	Stojanovic, Vladeta, Hagedorn, Benjamin, Trapp, Matthias and DÃ¶llner, JÃ¼rgen
year	2020
title	Ontology-Driven Analytics for Indoor Point Clouds
source	D. Holzer, W. Nakapan, A. Globa, I. Koh (eds.), RE: Anthropocene, Design in the Age of Humans - Proceedings of the 25th CAADRIA Conference - Volume 2, Chulalongkorn University, Bangkok, Thailand, 5-6 August 2020, pp. 537-546
doi	https://doi.org/10.52842/conf.caadria.2020.2.537
summary	Automated processing, semantic enrichment and visual analytics methods for point clouds are often use-case specific for a given domain (e.g, for Facility Management (FM) applications). Currently, this means that applicable processing techniques, semantics and visual analytics methods need to be selected, generated or implemented by human domain experts, which is an error-prone, subjective and non-interoperable process. An ontology-driven analytics approach can be used to solve this problem by creating and maintaining a Knowledge Base, and utilizing an ontology for automatically suggesting optimal selection of processing and analytics techniques for point clouds. We present an approach of an ontology-driven analytics concept and system design, which supports smart representation, exploration, and processing of indoor point clouds. We present and provide an overview of high-level concept and architecture for such a system, along with related key technologies and approaches based on previously published case studies. We also describe key requirements for system components, and discuss the feasibility of their implementation within a Service-Oriented Architecture (SOA).
keywords	Knowledge Base; Point Clouds; Semantic Enrichment; Service-Oriented Architecture; Ontology
series	CAADRIA
email
last changed	2022/06/07 07:56

_id	caadria2020_099
id	caadria2020_099
authors	Tu, Chun Man and Hou, June Hao
year	2020
title	After Abstraction, Before Figuration - Exploring the Potential Development of Form Re-topology and Evolution Reapplication with Three-dimensional Point Cloud Model Generation Logic.
source	D. Holzer, W. Nakapan, A. Globa, I. Koh (eds.), RE: Anthropocene, Design in the Age of Humans - Proceedings of the 25th CAADRIA Conference - Volume 2, Chulalongkorn University, Bangkok, Thailand, 5-6 August 2020, pp. 517-526
doi	https://doi.org/10.52842/conf.caadria.2020.2.517
summary	In the era of three-dimensional (3D) informatics, the 3D point cloud modeling algorithm has the potential to further develop. In this study, we attempt to eliminate the limitations of the traditional reverse modeling method and directly turn point cloud data into the material for innovative architectural design by integrating 3D point cloud modeling into the CAD/CAM platform(Rhino/Grasshopper) most widely used by parametric designers. In this way, the randomly ordered point cloud model can be regenerated and reordered according to the designer's requirements. In addition, point cloud data can be spatially segmented and morphologically evolved according to the designer's preferences to construct a 3D model with higher efficiency and more dynamic real-time adjustment compared with the triangular mesh model. Moreover, when a computer vision technique is integrated into the point cloud design process, the point cloud model can be further used to more efficiently achieve rapid visualization, artisticization, and form adjustment. Therefore, point cloud modeling can not only be applied to the spatial structure presentation of building information modeling(BIM) but also can provide further opportunities for creative architectural design.
keywords	Three-dimensional Point-cloud Model; Computer Vision; Point Set Registration; Topology Optimization; Regeneration
series	CAADRIA
email
last changed	2022/06/07 07:57

_id	sigradi2020_425
id	sigradi2020_425
authors	Vizioli, Simone Helena Tanoue; Adami, Andrea; Hiilesmaa, Laura; Carvalho, Leonardo Chieppe
year	2020
title	Comparative study of the photogrammetry process in different hardware
source	SIGraDi 2020 [Proceedings of the 24th Conference of the Iberoamerican Society of Digital Graphics - ISSN: 2318-6968] Online Conference 18 - 20 November 2020, pp. 425-432
summary	Photogrammetry is not a new technology, but its use, nowadays is being potentiated either in urban projects, architecture, restoration of historical heritage and archaeological documentation, among others. In this research, the use of photogrammetry as an important tool in the preservation and documentation of cultural heritage will be addressed. The specific objective of the article is to analyze the photogrammetry modeling process: processing time, resolution quality and generated products, in 2 different hardware; highlighting its potentialities and deficiencies. The object of study is the Casa do Pinhal, an important historical heritage for the identity and memory of the city of Sao Carlos (SP-Brazil).
keywords	Photogrammetry, 3D Modeling, Metashape, Heritage education, Point cloud
series	SIGraDi
email
last changed	2021/07/16 11:49

_id	acadia20_110
id	acadia20_110
authors	Zhang, Mengni; Dewey, Clara; Kalantari, Saleh
year	2020
title	Dynamic Anthropometric Modeling Interface
source	ACADIA 2020: Distributed Proximities / Volume I: Technical Papers [Proceedings of the 40th Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-95213-0]. Online and Global. 24-30 October 2020. edited by B. Slocum, V. Ago, S. Doyle, A. Marcus, M. Yablonina, and M. del Campo. 110-119.
doi	https://doi.org/10.52842/conf.acadia.2020.1.110
summary	In this paper, we propose a Kinect-based Dynamic Anthropometric Modeling Interface (DAMI), built in Rhinoceros with Grasshopper for patient room layout optimization and nurse posture evaluations. Anthropometry is an important field that studies human body measurements to help designers improve product ergonomics and reduce negative health consequences such as musculoskeletal disorders (MSDs). Unlike existing anthropometric tools, which rely on generic human body datasets and static posture models, DAMI tracks and records user postures in real time, creating custom 3D body movement models that are typically absent in current space-planning practices. A generic hospital patient room, which contains complex and ergonomically demanding activities for nurses, was selected as an initial testing environment. We will explain the project background, the methods used to develop DAMI, and demonstrate its capabilities. There are two main goals DAMI aims to achieve. First, as a generative tool, it will reconstruct dynamic body point cloud models, which will be used as input for optimizing room layout during a project’s schematic design phase. Second, as an evaluation tool, by encoding and visualizing the Rapid Entire Body Assessment (REBA) scores, DAMI will illustrate the spatiotemporal relationship between nurse postures and the built environment during a project’s construction phase or post occupancy evaluation. We envision a distributed system of Kinect sensors to be embedded in various hospital rooms to help architects, planners, and facility managers improve nurse work experiences through better space planning.
series	ACADIA
type	paper
email
last changed	2023/10/22 12:06

_id	caadria2020_398
id	caadria2020_398
authors	Tseng, Li-Min and Hou, June-Hao
year	2020
title	Representation of Sound in 3D
source	D. Holzer, W. Nakapan, A. Globa, I. Koh (eds.), RE: Anthropocene, Design in the Age of Humans - Proceedings of the 25th CAADRIA Conference - Volume 1, Chulalongkorn University, Bangkok, Thailand, 5-6 August 2020, pp. 609-618
doi	https://doi.org/10.52842/conf.caadria.2020.1.609
summary	This study is based on Chladni figures and tries to spatially extend its representation of sound. The current Chladni figures only see parts of the sound. There should be more spatial representation of sounds because they are transmitted in space. This study explores how to capture and reconstruct invisible sound information to create three-dimensional forms. A series of steps are taken to record Chladni figures of different frequencies and decibels. Pure Data is used to generate sounds. The Chladni figures are captured in Grasshopper and converted into point clouds. These point clouds are processed by using different algorithms to produce layers of superimposed state from which 3D forms of sound can be generated and fabricated. Through the proposed methods of processing and representation, sound not only stays at the level of hearing, but can also be seen, touched, and reinterpreted spatially. With the spatial forms of sound, viewers no longer perceive sound through single but multiple states. This can help us comprehend sound in a vast variety of ways.
keywords	Sound visualization; Form-finding; Spatial-temporal; Chladni figures; Cymatics
series	CAADRIA
email
last changed	2022/06/07 07:57

_id	acadia20_228
id	acadia20_228
authors	Alawadhi, Mohammad; Yan, Wei
year	2020
title	BIM Hyperreality
source	ACADIA 2020: Distributed Proximities / Volume I: Technical Papers [Proceedings of the 40th Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-95213-0]. Online and Global. 24-30 October 2020. edited by B. Slocum, V. Ago, S. Doyle, A. Marcus, M. Yablonina, and M. del Campo. 228-236.
doi	https://doi.org/10.52842/conf.acadia.2020.1.228
summary	Deep learning is expected to offer new opportunities and a new paradigm for the field of architecture. One such opportunity is teaching neural networks to visually understand architectural elements from the built environment. However, the availability of large training datasets is one of the biggest limitations of neural networks. Also, the vast majority of training data for visual recognition tasks is annotated by humans. In order to resolve this bottleneck, we present a concept of a hybrid system—using both building information modeling (BIM) and hyperrealistic (photorealistic) rendering—to synthesize datasets for training a neural network for building object recognition in photos. For generating our training dataset, BIMrAI, we used an existing BIM model and a corresponding photorealistically rendered model of the same building. We created methods for using renderings to train a deep learning model, trained a generative adversarial network (GAN) model using these methods, and tested the output model on real-world photos. For the specific case study presented in this paper, our results show that a neural network trained with synthetic data (i.e., photorealistic renderings and BIM-based semantic labels) can be used to identify building objects from photos without using photos in the training data. Future work can enhance the presented methods using available BIM models and renderings for more generalized mapping and description of photographed built environments.
series	ACADIA
type	paper
email
last changed	2023/10/22 12:06

_id	ecaade2020_017
id	ecaade2020_017
authors	Chan, Yick Hin Edwin and Spaeth, A. Benjamin
year	2020
title	Architectural Visualisation with Conditional Generative Adversarial Networks (cGAN). - What machines read in architectural sketches.
source	Werner, L and Koering, D (eds.), Anthropologic: Architecture and Fabrication in the cognitive age - Proceedings of the 38th eCAADe Conference - Volume 2, TU Berlin, Berlin, Germany, 16-18 September 2020, pp. 299-308
doi	https://doi.org/10.52842/conf.ecaade.2020.2.299
summary	As a form of visual reasoning, sketching is a human cognitive activity instrumental to architectural design. In the process of sketching, abstract sketches invoke new mental imageries and subsequently lead to new sketches. This iterative transformation is repeated until the final design emerges. Artificial Intelligence and Deep Neural Networks have been developed to imitate human cognitive processes. Amongst these networks, the Conditional Generative Adversarial Network (cGAN) has been developed for image-to-image translation and is able to generate realistic images from abstract sketches. To mimic the cyclic process of abstracting and imaging in architectural concept design, a Cyclic-cGAN that consists of two cGANs is proposed in this paper. The first cGAN transforms sketches to images, while the second from images to sketches. The training of the Cyclic-cGAN is presented and its performance illustrated by using two sketches from well-known architects, and two from architecture students. The results show that the proposed Cyclic-cGAN can emulate architects' mode of visual reasoning through sketching. This novel approach of utilising deep neural networks may open the door for further development of Artificial Intelligence in assisting architects in conceptual design.
keywords	visual cognition; design computation; machine learning; artificial intelligence
series	eCAADe
email
last changed	2022/06/07 07:55

_id	caadria2020_446
id	caadria2020_446
authors	Cho, Dahngyu, Kim, Jinsung, Shin, Eunseo, Choi, Jungsik and Lee, Jin-Kook
year	2020
title	Recognizing Architectural Objects in Floor-plan Drawings Using Deep-learning Style-transfer Algorithms
source	D. Holzer, W. Nakapan, A. Globa, I. Koh (eds.), RE: Anthropocene, Design in the Age of Humans - Proceedings of the 25th CAADRIA Conference - Volume 2, Chulalongkorn University, Bangkok, Thailand, 5-6 August 2020, pp. 717-725
doi	https://doi.org/10.52842/conf.caadria.2020.2.717
summary	This paper describes an approach of recognizing floor plans by assorting essential objects of the plan using deep-learning based style transfer algorithms. Previously, the recognition of floor plans in the design and remodeling phase was labor-intensive, requiring expert-dependent and manual interpretation. For a computer to take in the imaged architectural plan information, the symbols in the plan must be understood. However, the computer has difficulty in extracting information directly from the preexisting plans due to the different conditions of the plans. The goal is to change the preexisting plans to an integrated format to improve the readability by transferring their style into a comprehensible way using Conditional Generative Adversarial Networks (cGAN). About 100-floor plans were used for the dataset which was previously constructed by the Ministry of Land, Infrastructure, and Transport of Korea. The proposed approach has such two steps: (1) to define the important objects contained in the floor plan which needs to be extracted and (2) to use the defined objects as training input data for the cGAN style transfer model. In this paper, wall, door, and window objects were selected as the target for extraction. The preexisting floor plans would be segmented into each part, altered into a consistent format which would then contribute to automatically extracting information for further utilization.
keywords	Architectural objects; floor plan recognition; deep-learning; style-transfer
series	CAADRIA
email
last changed	2022/06/07 07:56

_id	acadia20_272
id	acadia20_272
authors	del Campo, Matias; Carlson, Alexandra; Manninger, Sandra
year	2020
title	How Machines Learn to Plan
source	ACADIA 2020: Distributed Proximities / Volume I: Technical Papers [Proceedings of the 40th Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-95213-0]. Online and Global. 24-30 October 2020. edited by B. Slocum, V. Ago, S. Doyle, A. Marcus, M. Yablonina, and M. del Campo. 272-281.
doi	https://doi.org/10.52842/conf.acadia.2020.1.272
summary	This paper strives to interrogate the abilities of machine vision techniques based on a family of deep neural networks, called generative adversarial neural networks (GANs), to devise alternative planning solutions. The basis for these processes is a large database of existing planning solutions. For the experimental setup of this paper, these plans were divided into two separate learning classes: Modern and Baroque. The proposed algorithmic technique leverages the large amount of structural and symbolic information that is inherent to the design of planning solutions throughout history to generate novel unseen plans. In this area of inquiry, aspects of culture such as creativity, agency, and authorship are discussed, as neural networks can conceive solutions currently alien to designers. These can range from alien morphologies to advanced programmatic solutions. This paper is primarily interested in interrogating the second existing but uncharted territory.
series	ACADIA
type	paper
email
last changed	2023/10/22 12:06

For more results click below: