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 537

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

_id	ijac202220216
id	ijac202220216
authors	Keyvanfar, Ali; Arezou Shafaghat; Muhamad SF Rosley
year	2022
title	Performance comparison analysis of 3D reconstruction modeling software in construction site visualization and mapping
source	International Journal of Architectural Computing 2022, Vol. 20 - no. 2, pp. 453–475
summary	Unmanned aerial vehicle (UAV) technology has overcome the limitations of conventional construction management methods using advanced and automated visualization and 3D reconstruction modeling techniques. Although the mapping techniques and reconstruction modeling software can generate real-time and high-resolution descriptive textural, physical, and spatial data, they may fail to develop an accurate and complete 3D model of the construction site. To generate a quality 3D reconstruction model, the construction manager must optimize the trade-offs among three major software-selection factors: functionalities, technical capabilities, and the system hardware specifications. These factors directly affect the robust 3D reconstruction model of the construction site and objects. Accordingly, the purpose of this research was to apply nine well-established 3D reconstruction modeling software tools (DroneDeploy, COLMAP, 3DF+Zephyr, Autodesk Recap, LiMapper, PhotoModeler, 3D Survey, AgiSoft Photoscan, and Pix4D Mapper) and compare their performances and reliabilities in generating complete 3D models. The research was conducted in an eco-home building at the University of Technology, Malaysia. A series of regression analyses were conducted to compare the performances of the selected 3D reconstruction modeling software in alignment and registration, distance computing, geometric measurement, and plugin execution. Regression analysis determined that among the software programs, LiMapper had the strongest positive linear correlation with the ground truth model. Furthermore, the correlation analysis showed a statistically significant p-value for all software, except for 3D Survey. In addition, the research found that Autodesk Recap generated the most-robust and highest-quality dense point clouds. DroneDeploy can create an accurate point cloud and triangulation without using many points as required by COLMAP and LiMapper. It was concluded that most of the software is robustly, positively, and linearly correlated with the corresponding ground truth model. In the future, other factors involving software selection should be studied, such as vendor-related, user-related, and automation factors.
keywords	Construction site visualization, unmanned aerial vehicle, photogrammetry, 3D reconstruction modeling, multi-view-stereopsis, structure-from-motion, ANOVA and regression analysis
series	journal
last changed	2024/04/17 14:29

_id	sigradi2022_168
id	sigradi2022_168
authors	Koh, Immanuel
year	2022
title	Palette2Interior Architecture: From Syntactic and Semantic Colour Palettes to Generative Interiors with Deep Learning
source	Herrera, PC, Dreifuss-Serrano, C, Gómez, P, Arris-Calderon, LF, Critical Appropriations - Proceedings of the XXVI Conference of the Iberoamerican Society of Digital Graphics (SIGraDi 2022), Universidad Peruana de Ciencias Aplicadas, Lima, 7-11 November 2022 , pp. 187–198
summary	Colour palettes have long played a significant role in not only capturing design ambience (e.g., as mood boards), but more significantly, in translating an abstract intuition into an explicit ordering mechanism for design representation and synthesis, whether it is in the discipline of graphic design, interior design or architectural design. Might this difficult process of design synthesis from a low-dimensional colour input domain to a high-dimensional spatial design output domain be computationally mapped? Using today’s generative adversarial networks (GANs), the paper aims to investigate this plausibility, and in doing so, hoping to envision an AI-augmented design workflow and tooling. Newly-created datasets are made procedurally and used to train three different types of deep learning models in the specific context of generating living room interior layouts. The results suggest that a combination of syntactic and semantic generative processes is necessary for a critical appropriation of such AI models
keywords	Machine Learning, Artificial Intelligence, Deep Neural Networks, Colour Palette, Interior Design
series	SIGraDi
email
last changed	2023/05/16 16:55

_id	caadria2022_177
id	caadria2022_177
authors	Pan, Yongjie and Zhang, Tong
year	2022
title	Outdoor Thermal Environment Assessment of Existing Residential Areas Supported by UAV Thermal Infrared and 3D Reconstruction Technology
source	Jeroen van Ameijde, Nicole Gardner, Kyung Hoon Hyun, Dan Luo, Urvi Sheth (eds.), POST-CARBON - Proceedings of the 27th CAADRIA Conference, Sydney, 9-15 April 2022, pp. 729-738
doi	https://doi.org/10.52842/conf.caadria.2022.2.729
summary	The underlying surface temperature is an effective evaluation index to study the urban micro-scale thermal environment. For surface temperature acquisition, the thermal infrared camera mounted on a unmanned aerial vehicle (UAV) can reduce field work intensity, improve data collection efficiency, and ensure high accuracy at low cost. In order to convert the 2D thermal image into a more intuitive 3D thermal model, the UAV-based thermal infrared 3D reconstruction is adopted. The key element of thermal infrared 3D model reconstruction lies in the processing of thermal infrared images with low resolution and different temperature scales. In order to improve the quality of the final thermal 3D model, this paper proposes the reconstruction of the detailed 3D mesh using visible images (higher resolution), and map then mapping thermal textures onto the mesh using thermal images (low resolution). In addition, absolute temperature values are extracted from thermal images with different temperature ranges to ensure consistence between color and temperature values in the reconstructed thermal 3D model. The thermal 3D model generated for an existing residential area in Nanjing successfully displays the temperature distribution of the underlying surface and provides a valuable basis for outdoor thermal environment assessment.
keywords	Thermal image, UAV, 3D reconstruction, Residential outdoor space, Underlying surface temperature, SDG 3, SDG 11
series	CAADRIA
email
last changed	2022/07/22 07:34

_id	ecaade2022_217
id	ecaade2022_217
authors	Panagiotidou, Vasiliki and Koerner, Andreas
year	2022
title	From Intricate to Coarse and Back - A voxel-based workflow to approximate high-res geometries for digital environmental simulations
source	Pak, B, Wurzer, G and Stouffs, R (eds.), Co-creating the Future: Inclusion in and through Design - Proceedings of the 40th Conference on Education and Research in Computer Aided Architectural Design in Europe (eCAADe 2022) - Volume 1, Ghent, 13-16 September 2022, pp. 491–500
doi	https://doi.org/10.52842/conf.ecaade.2022.1.491
summary	Digital environmental simulations can present a computational bottleneck concerning the complexity of geometry. Therefore, a series of workarounds, ranging from cloud-based solutions to machine learning simulations as surrogate simulations are conventionally applied in practice. Concurrently, contemporary advances in procedural modelling in architecture result in design concepts with high polygon counts. This leads to an ever- increasing resolution discrepancy between design and analysis models. Responding to this problem, this research presents a step-by-step approximation workflow for handling and transferring high-resolution geometries between procedural modelling and environmental simulation software. The workflow is intended to allow designers to quickly assess a design’s interaction with environmental parameters such as airflow and solar radiation and further articulate them. A controllable voxelization procedure is applied to approximate the original geometry and therefore reduce the resolution. Controllable in this context refers to the user’s ability to locally adjust the voxel resolution to fit design needs. After export and simulation, 3d results are imported back into the design environment. The colour properties are re-mapped onto the original high- resolution geometry following a weighted proximity technique. The developed data transfer pipeline allows designers to integrate environmental analysis during initial design steps, which is essential for accessibility in the design profession. This can help to environmentally inform generative designs as well as to make simulation workflows more accessible when working with a wider range of geometries. In this, it reduces the perceived discrepancy between the concept and simulation model. This eases the use and allows a wider audience of users to develop co-creation processes between computation, architecture, and environment.
keywords	Simulation, Accessibility, Computation, Environmental Data, Workflow
series	eCAADe
email
last changed	2024/04/22 07:10

_id	ecaade2022_264
id	ecaade2022_264
authors	Sanatani, Rohit Priyadarshi
year	2022
title	Democratizing Urban Data - A smartphone-based framework for rapid cataloging of geolocated street-level imagery and visual content analysis
source	Pak, B, Wurzer, G and Stouffs, R (eds.), Co-creating the Future: Inclusion in and through Design - Proceedings of the 40th Conference on Education and Research in Computer Aided Architectural Design in Europe (eCAADe 2022) - Volume 1, Ghent, 13-16 September 2022, pp. 511–516
doi	https://doi.org/10.52842/conf.ecaade.2022.1.511
summary	The commercial availability of high-resolution street view imagery, most notably Google Street View, has led to its widespread use in urban analytics research over the past couple of years. Recent developments in computer vision, most notably semantic segmentation and object detection, have made it possible to extract and map the visual features of streetscapes (such as buildings, automobiles, green cover, pedestrians etc.) using geo-located street level photographs. However, the absence of such detailed imagery in many parts of the world stands as a significant deterrent to these research methodologies. A majority of countries in Africa, the Middle East, as well as some parts of Asia currently have limited coverage by street view image providers. The cost component and equipment involved in manual data collection stands as a barrier to accessible urban visual data. This paper demonstrates a quick and inexpensive smartphone-based framework for rapid and inexpensive collection and cataloging of geolocated street-level imagery. The user walks/drives down the streets to be mapped with a smartphone, as a first-person egocentric hyper-lapse video is recorded with a fixed frame interval, along with location information for the path taken. The video frames are then automatically extracted, geo-referenced and stored in a readily retrievable format. This data can then easily be used for urban feature extraction through computer vision workflows. For demonstration, imagery has been cataloged for a ~1.5 sq.km urban area in New Delhi, and then processed through a semantic segmentation workflow for visual feature mapping. It is hoped that this framework plays a role in democratizing access to street level data for students and researchers regardless of national boundaries.
keywords	Street View Imagery, Democratizing Data, Hyperlapse Photography, Smartphone, Urban Analytics
series	eCAADe
email
last changed	2024/04/22 07:10

_id	cdrf2022_371
id	cdrf2022_371
authors	Viktória Sándor, Mathias Bank, Kristina Schinegger, and Stefan Rutzinger
year	2022
title	Collapsing Complexities: Encoding Multidimensional Architecture Models into Images
source	Proceedings of the 2022 DigitalFUTURES The 4st International Conference on Computational Design and Robotic Fabrication (CDRF 2022)
doi	https://doi.org/https://doi.org/10.1007/978-981-19-8637-6_32
summary	The paper details a 3D to 2D encoding method, which can store complex digital 3D models of architecture within a single image. The proposed encoding works in combination with a point cloud notation and a sequential slicing operation where each slice of points is stored as a single row of pixels in the UV space of a 1024?×?1024 image. The performance of the notation system is compared between a StyleGan2 and existing image editing methods and evaluated through the production of new 3D models of houses with material attributes. The uncovered findings maintain the relatively high level of detail stored through the encoding while allowing for innovative ways of form-finding—producing new and unseen 3d models of architectural houses.
series	cdrf
email
last changed	2024/05/29 14:03

_id	cdrf2022_326
id	cdrf2022_326
authors	Zidong Liu, Yan Li, and Xiao Xiao
year	2022
title	Predicting the Vitality of Stores Along the Street Based on Business Type Sequence via Recurrent Neural Network
source	Proceedings of the 2022 DigitalFUTURES The 4st International Conference on Computational Design and Robotic Fabrication (CDRF 2022)
doi	https://doi.org/https://doi.org/10.1007/978-981-19-8637-6_29
summary	The rational planning of store types and locations to maximize street vitality is essential in real estate planning. Traditional business planning relies heavily on the subjective experience of developers. Currently, developers have access to low-resolution urban data to support their decision making, and researchers have done much image-based machine learning research from the scale of urban texture. However, there is still a lack of research on the functional layout with shop-level accuracy. This paper uses a sequence-based neural network (RNN) to explore the relationship between the sequence of store types along a street and its commercial vitality. Currently, the use of RNNs in the architectural and urban fields is very rare. We use customer review data of 80streets from O2O platforms to represent the store vitality degree. In the machine learning model, the input is the sequence of store types on the street, and the output is the corresponding sequence of business vitality indexes. After training and evaluation, the model was shown to have acceptable accuracy. We further combined this evaluation model with a genetic algorithm to develop a business planning optimization tool to maximize the overall street business value, thus guiding real estate business planning at a high resolution.
series	cdrf
email
last changed	2024/05/29 14:03

_id	caadria2022_336
id	caadria2022_336
authors	Araujo, Goncalo, Santos, Luis, Leitao, Antonioand Gomes, Ricardo
year	2022
title	AD-Based Surrogate Models for Simulation and Optimization of Large Urban Areas
source	Jeroen van Ameijde, Nicole Gardner, Kyung Hoon Hyun, Dan Luo, Urvi Sheth (eds.), POST-CARBON - Proceedings of the 27th CAADRIA Conference, Sydney, 9-15 April 2022, pp. 689-698
doi	https://doi.org/10.52842/conf.caadria.2022.2.689
summary	Urban Building Energy Model (UBEM) approaches help analyze the energy performance of urban areas and predict the impact of different retrofit strategies. However, UBEM approaches require a high level of expertise and entail time-consuming simulations. These limitations hinder their successful application in designing and planning urban areas and supporting the city policy-making sector. Hence, it is necessary to investigate alternatives that are easy-to-use, automated, and fast. Surrogate models have been recently used to address UBEM limitations; however, they are case-specific and only work properly within specific parameter boundaries. We propose a new surrogate modeling approach to predict the energy performance of urban areas by integrating Algorithmic Design, UBEM, and Machine Learning. Our approach can automatically model and simulate thousands of building archetypes and create a broad surrogate model capable of quickly predicting annual energy profiles of large urban areas. We evaluated our approach by applying it to a case study located in Lisbon, Portugal, where we compare its use in model-based optimization routines against conventional UBEM approaches. Results show that our approach delivers predictions with acceptable accuracy at a much faster rate.
keywords	urban building energy modelling, algorithmic design, machine learning in Architecture, optimization of urban areas, SDG 7, SDG 12, SDG 13
series	CAADRIA
email
last changed	2022/07/22 07:34

_id	caadria2022_45
id	caadria2022_45
authors	Boim, Anna, Dortheimer, Jonathan and Sprecher, Aaron
year	2022
title	A Machine-Learning Approach to Urban Design Interventions In Non-Planned Settlements
source	Jeroen van Ameijde, Nicole Gardner, Kyung Hoon Hyun, Dan Luo, Urvi Sheth (eds.), POST-CARBON - Proceedings of the 27th CAADRIA Conference, Sydney, 9-15 April 2022, pp. 223-232
doi	https://doi.org/10.52842/conf.caadria.2022.1.223
summary	This study presents generative adversarial networks (GANs), a machine-learning technique that can be used as an urban design tool capable of learning and reproducing complex patterns that express the unique spatial qualities of non-planned settlements. We report preliminary experimental results of training and testing GAN models on different datasets of urban patterns. The results reveal that machine learning models can generate development alternatives with high morphological resemblance to the original urban fabric based on the suggested training process. This study contributes a methodological framework that has the potential to generate development alternatives sensitive to the local practices, thereby promoting preservation of traditional knowledge and cultural sustainability.
keywords	Non-planned settlements, Cultural Sustainability, Machine Learning, Generative Adversarial Networks, SDG 11
series	CAADRIA
email
last changed	2022/07/22 07:34

_id	sigradi2022_268
id	sigradi2022_268
authors	Fernandez Gonzalez, Alberto
year	2022
title	A neo-postnatural high resolution aesthetic by Cellular Architecture
source	Herrera, PC, Dreifuss-Serrano, C, Gómez, P, Arris-Calderon, LF, Critical Appropriations - Proceedings of the XXVI Conference of the Iberoamerican Society of Digital Graphics (SIGraDi 2022), Universidad Peruana de Ciencias Aplicadas, Lima, 7-11 November 2022 , pp. 163–172
summary	The level of detail in architecture is part of our legacy since our discipline has existed as a catalogue of parts, linking art and nature by using concepts of imitation, selection, neatness, and ornament. This research expands the idea of “high resolution” in architecture as a step forward in defining detail and ornament as a way of finally merging the structural and functional dimension of space with the detail and ornamental dimension of a design proposal. In that framework, this work-in-progress research emerges as an opportunity to use Cellular Automata (CA) and its principles to create coherence between different scales in the inhabitable space from a bottom-up approach. CA principles work in both computational levels (as logical machines) and a biological simulator (artificial life), enabling us to cross borders between natural and unnatural in a multi-scale design approach.
keywords	Generative Design, Detail, High Resolution, Cellular Automata, Multi-scalar
series	SIGraDi
email
last changed	2023/05/16 16:55

_id	caadria2022_239
id	caadria2022_239
authors	Huang, Chenyu, Zhang, Gengjia, Yin, Minggang and Yao, Jiawei
year	2022
title	Energy-driven Intelligent Generative Urban Design, Based on Deep Reinforcement Learning Method With a Nested Deep Q-R Network
source	Jeroen van Ameijde, Nicole Gardner, Kyung Hoon Hyun, Dan Luo, Urvi Sheth (eds.), POST-CARBON - Proceedings of the 27th CAADRIA Conference, Sydney, 9-15 April 2022, pp. 233-242
doi	https://doi.org/10.52842/conf.caadria.2022.1.233
summary	To attain "carbon neutrality," lowering urban energy use and increasing the use of renewable resources have become critical concerns for urban planning and architectural design. Traditional energy consumption evaluation tools have a high operational threshold, requiring specific parameter settings and cross-disciplinary knowledge of building physics. As a result, it is difficult for architects to manage energy issues through 'trial and error' in the design process. The purpose of this study is to develop an automated workflow capable of providing urban configurations that minimizing the energy use while maximizing rooftop photovoltaic power potential. Based on shape grammar, parametric meta models of three different urban forms were developed and batch simulated for its energy performance. Deep reinforcement learning (DRL) is introduced to find the optimal solution of the urban geometry. A neural network was created to fit a real-time mapping of urban form indicators to energy performance and was utilized to predict reward for the DRL process, namely a Deep R-Network, while nested within a Deep Q-Network. The workflow proposed in this paper promotes efficiency in optimizing the energy performance of solutions in the early stages of design, as well as facilitating a collaborative design process with human-machine interaction.
keywords	energy-driven urban design, intelligent generative design, rooftop photovoltaic power, deep reinforcement learning, SDG 11, SDG 12
series	CAADRIA
email
last changed	2022/07/22 07:34

_id	architectural_intelligence2022_5
id	architectural_intelligence2022_5
authors	Jiading Zhong, Jianlin Liu, Yongling Zhao, Jianlei Niu & Jan Carmeliet
year	2022
title	Recent advances in modeling turbulent wind flow at pedestrian-level in the built environment
source	Architectural Intelligence Journal
doi	https://doi.org/https://doi.org/10.1007/s44223-022-00008-7
summary	Pressing problems in urban ventilation and thermal comfort affecting pedestrians related to current urban development and densification are increasingly dealt with from the perspective of climate change adaptation strategies. In recent research efforts, the prime objective is to accurately assess pedestrian-level wind (PLW) environments by using different simulation approaches that have reasonable computational time. This review aims to provide insights into the most recent PLW studies that use both established and data-driven simulation approaches during the last 5 years, covering 215 articles using computational fluid dynamics (CFD) and typical data-driven models. We observe that steady-state Reynolds-averaged Navier-Stokes (SRANS) simulations are still the most dominantly used approach. Due to the model uncertainty embedded in the SRANS approach, a sensitivity test is recommended as a remedial measure for using SRANS. Another noted thriving trend is conducting unsteady-state simulations using high-efficiency methods. Specifically, both the massively parallelized large-eddy simulation (LES) and hybrid LES-RANS offer high computational efficiency and accuracy. While data-driven models are in general believed to be more computationally efficient in predicting PLW dynamics, they in fact still call for substantial computational resources and efforts if the time for development, training and validation of a data-driven model is taken into account. The synthesized understanding of these modeling approaches is expected to facilitate the choosing of proper simulation approaches for PLW environment studies, to ultimately serving urban planning and building designs with respect to pedestrian comfort and urban ventilation assessment.
series	Architectural Intelligence
email
last changed	2025/01/09 15:00

_id	caadria2022_286
id	caadria2022_286
authors	Khean, Nariddh, During, Serjoscha, Chronis, Angelos, Konig, Reinhard and Haeusler, Matthias Hank
year	2022
title	An Assessment of Tool Interoperability and its Effect on Technological Uptake for Urban Microclimate Prediction with Deep Learning Models
source	Jeroen van Ameijde, Nicole Gardner, Kyung Hoon Hyun, Dan Luo, Urvi Sheth (eds.), POST-CARBON - Proceedings of the 27th CAADRIA Conference, Sydney, 9-15 April 2022, pp. 273-282
doi	https://doi.org/10.52842/conf.caadria.2022.1.273
summary	The benefits of deep learning (DL) models often overshadow the high costs associated with training them. Especially when the intention of the resultant model is a more climate resilient built environment, overlooking these costs are borderline hypocritical. However, the DL models that model natural phenomena‚conventionally simulated through predictable mathematical modelling‚don't succumb to the costly pitfalls of retraining when a model's predictions diverge from reality over time. Thus, the focus of this research will be on the application of DL models in urban microclimate simulations based on computational fluid dynamics. When applied, predicting wind factors through DL, rather than arduously simulating, can offer orders of magnitude of improved computational speed and costs. However, despite the plethora of research conducted on the training of such models, there is comparatively little work done on deploying them. This research posits: to truly use DL for climate resilience, it is not enough to simply train models, but also to deploy them in an environment conducive of rapid uptake with minimal barrier to entry. Thus, this research develops a Grasshopper plugin that offers planners and architects the benefits gained from DL. The outcomes of this research will be a tangible tool that practitioners can immediately use, toward making effectual change.
keywords	Deep Learning, Technological Adoption, Fluid Dynamics, Urban Microclimate Simulation, Grasshopper, SDG 11
series	CAADRIA
email
last changed	2022/07/22 07:34

_id	ijac202220303
id	ijac202220303
authors	Kirdar, Gulce; Gulen Cagdas
year	2022
title	A decision support model to evaluate liveability in the context of urban vibrancy
source	International Journal of Architectural Computing 2022, Vol. 20 - no. 3, pp. 528–552
summary	Liveability can be accepted as an umbrella term covering all the factors that make a place to live. We recognize the versatility of urban liveability and focus on the vibrancy aspect. Regarding the literature, we compile variables affecting urban liveability under the economic, image, and use value of place. This article aims to present a data-driven decision support system to evaluate different dimensions of vibrancy-focused liveability. We adopt a knowledge discovery process to handle the complexity of the liveability concept. This study develops a conditional-based relationship network of vibrancy parameters through the Bayesian Belief Network (BBN). Then, we assess the BBN’s correlations with statistics and causal relations with the survey in this study.These results mostly agree with the findings of the relevant literature. The economic value results show that the high density, diversity and accessibility add a premium to the land value of properties. The use value results also demonstrate that the diversity and density of activities, cultural attributes, and high accessibility support place attractiveness. The selected streetscape variables improve image value, except for building enclosure and condition. The study has the potential for urban planners to vitalize neighborhoods by considering urban activities and urban physical attributes
keywords	liveability, vibrancy, knowledge discovery process, big data, locative data, Bayesian belief network
series	journal
last changed	2024/04/17 14:29

_id	ijac202220109
id	ijac202220109
authors	Ortner, F. Peter; Jing Zhi Tay
year	2022
title	Resilient by design: Informing pandemic-safe building redesign with computational models of resident congestion
source	International Journal of Architectural Computing 2022, Vol. 20 - no. 1, pp. 129–144
summary	This paper describes a computational design-support tool created in response to safe-distancing measures enforced during the COVID-19 pandemic. The tool was developed for a specific use case: understanding congestion in crowded migrant worker dormitories that experienced high rates of COVID-19 transmission in 2020. Building from agent-based and network-based computational simulations, the tool presents a hybrid method for simulating building resident movements based on known or pre-determined schedules and likely itineraries. This hybrid method affords the design tool a novel approach to simultaneous exploration of spatial and temporal design scenarios. The paper demonstrates the use of the tool on an anonymised case study of a high-density migrant worker dormitory, comparing results from a baseline configuration against design variations that modify dormitory physical configuration and schedule. Comparisons between the design scenarios provide evidence for reflections on pandemic-resilient design and operation strategies for dor- mitories. A conclusions section considers the extent to which the model and case study results are applicable to other dense institutional buildings and describes the paper’s contributions to general understanding of configurational and operational aspects of resilience in the built environment.
keywords	Design for resilience, evidence-based design, design support, agent-based model, schedule-based model, network analysis
series	journal
last changed	2024/04/17 14:29

_id	ijac202220308
id	ijac202220308
authors	Rodrigues, Ricardo C; Rovenir B Duarte
year	2022
title	Generating floor plans with deep learning: A cross-validation assessment over different dataset sizes
source	International Journal of Architectural Computing 2022, Vol. 20 - no. 3, pp. 630–644
summary	The advent of deep learning has enabled a series of opportunities; one of them is the ability to tackle subjective factors on the floor plan design and make predictions though spatial semantic maps. Nonetheless, the amount available of data grows exponentially on a daily basis, in this sense, this research seeks to investigate deep generative methods of floor plan design and its relationship between data volume, with training time, quality and diversity in the outputs; in other words, what is the amount of data required to rapidly train models that return optimal results. In our research, we used a variation of the Conditional Generative Adversarial Network algorithm, that is, Pix2pix, and a dataset of approximately 80 thousand images to train 10 models and evaluate their performance through a series of computational metrics. The results show that the potential of this data-driven method depends not only on the diversity of the training set but also on the linearity of the distribution; therefore, high-dimensional datasets did not achieve good results. It is also concluded that models trained on small sets of data (800 images) may return excellent results if given the correct training instructions (Hyperparameters), but the best baseline to this generative task is in the mid-term, using around 20 to 30 thousand images with a linear distribution. Finally, it is presented standard guidelines for dataset design, and the impact of data curation along the entire process
keywords	Dataset Reduction, Pix2pix, Artificial Intelligence, Deep Generative Models, GANs
series	journal
last changed	2024/04/17 14:30

_id	sigradi2022_253
id	sigradi2022_253
authors	Sanatani, Rohit Priyadarshi; Nagakura, Takehiko; Tsai, Daniel
year	2022
title	The Tourist’s Image of the City: A comparative analysis of visual features and textual themes of interest across three metropolises
source	Herrera, PC, Dreifuss-Serrano, C, Gómez, P, Arris-Calderon, LF, Critical Appropriations - Proceedings of the XXVI Conference of the Iberoamerican Society of Digital Graphics (SIGraDi 2022), Universidad Peruana de Ciencias Aplicadas, Lima, 7-11 November 2022 , pp. 89–100
summary	Tourist attractions play a major role in shaping ‘mental images’ of cities. The growing availability of urban big-data in recent years has opened up novel lines of inquiry into the nuances of urban imageability and sentiment. Drawing upon crowdsourced hybrid data in the form of both textual descriptions as well as photographs for 750 tourist attractions across Boston, Singapore and Sydney, this work compares the predominant themes of discussion and visual features of interest that shape tourist sentiment towards these cities. The study collects over 3500 user reviews and uses Latent Dirichlet Allocation (LDA) for the extraction of high-level topics of discussion. Object detection is also run on over 6000 photographs, and unsupervised clustering is carried out on extracted features to identify clusters of visual elements which capture tourist attention. The findings reinforce the popular identity of Boston as a city steeped in history, while strong perceptions of nature and greenery emerge from Singapore. Tourist interest in Sydney is dominated by specific anchors such as the Sydney Harbor Bridge.
keywords	Data Analytics, Urban Tourism, Topic Modeling, Sentiment Analysis, Unsupervised Clustering, Big Data
series	SIGraDi
email
last changed	2023/05/16 16:55

_id	acadia22_196
id	acadia22_196
authors	Sunshine, Gil
year	2022
title	Inventory
source	ACADIA 2022: Hybrids and Haecceities [Proceedings of the 42nd Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 979-8-9860805-8-1]. University of Pennsylvania Stuart Weitzman School of Design. 27-29 October 2022. edited by M. Akbarzadeh, D. Aviv, H. Jamelle, and R. Stuart-Smith. 196-207.
summary	Inventory offers an alternative to contemporary CAD software, where the gap between digital models and physical constraints is vast. Rather than abstract commands that project forth a not yet existing material condition, Inventory is based on digital representations of specific pieces of material and processes for fabricating assemblies of parts. By the very nature of their being digital, these representations are necessarily approximations of their physical counterparts. They inhabit the space between the low resolution of pure geometric abstraction and high resolution of physical phenomena, and therefore, might be called “medium resolution” (Sunshine 2022). Inventory uses game engine physics to embed simulations of physical constraints in the digital modeling process. Inventory is a software interface for making architecture in a medium resolution world.
series	ACADIA
type	paper
email
last changed	2024/02/06 14:00

_id	acadia22_524
id	acadia22_524
authors	Xiao, Jun; Liu, Yubo; Deng, Qiaoming
year	2022
title	High-Density Building Form Generation Considering Daylight Performance
source	ACADIA 2022: Hybrids and Haecceities [Proceedings of the 42nd Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 979-8-9860805-8-1]. University of Pennsylvania Stuart Weitzman School of Design. 27-29 October 2022. edited by M. Akbarzadeh, D. Aviv, H. Jamelle, and R. Stuart-Smith. 524-535.
summary	In this case, aiming for a high-density building designed with high-quality daylighting, this article develops a building form generation program for daylight performance optimization integrating Cellular Automata (CA) and Genetic Algorithm (GA), tools that can provide a global daylighting optimization through the balance of competition and concession of agents. The CA model provides randomness and structural restriction, while GA provide optimization and convergence by evaluating and selecting a series of CA models. The model applies designed CA rules on daylighting and a mathematic proxy model for daylight performance in GA fitness calculation.
series	ACADIA
type	paper
email
last changed	2024/02/06 14:04

_id	ecaade2022_273
id	ecaade2022_273
authors	Zhuang, Xinwei
year	2022
title	Rendering Sketches - Interactive rendering generation from sketches using conditional generative adversarial neural network
source	Pak, B, Wurzer, G and Stouffs, R (eds.), Co-creating the Future: Inclusion in and through Design - Proceedings of the 40th Conference on Education and Research in Computer Aided Architectural Design in Europe (eCAADe 2022) - Volume 1, Ghent, 13-16 September 2022, pp. 517–524
doi	https://doi.org/10.52842/conf.ecaade.2022.1.517
summary	Architects use sketches in the early design phase to organize and elaborate their initial ideas, and those initial sketches often support ideation for the final design. However, the sketches in the early design phase tend to be abstract and hard to interpret. Minimal prior works provide tools for quick visualization of the initial sketch. This study provides a scheme for architects and designers to generate preliminary renderings in the early design stage. In this study, we use conditional generative adversarial networks (cGAN) as the frame and introduces an updater network to the existing cGAN to support the iterative design process. A sketch serves as input to see the rendering and update the sketch based on the generated renderings by adding more resolution and details. The network is able to generate a reasonable rendering from the single-image network, and is able to update the renderings iteratively via the updater network. The dataset is collected from residential buildings exclusively, but the architectural categories can be expanded to other types of buildings in the future. Results show that the proposed scheme is able to provide reasonable renderings from sketches, and the generated rendering can be updated with a higher level of details within a second if the user provides a more detailed sketch. The contribution of this study includes introducing an updater network to the existing algorithm to enable iterative input and provides an alternative enhancement approach to the resolution of the generated image.
keywords	Computer Aided Design, Early Design Phase, Conditional Generative Adversarial Neural Network, Human Computer Interaction
series	eCAADe
email
last changed	2024/04/22 07:10

For more results click below: