Home
-
Computation & Design
Urban Cognition
MBTA Upzone+
Virtual Vitality
Dogs’ Journey
Boundary of Heterotopia
Heritage Extension
Architecture
Bamboo Dome
Professional Works
About Me
Home
Urban Cognition
MBTA Upzone+
Virtual Vitality
Dogs’ Journey
Boundary of Heterotopia
Heritage Extension
Architecture
Bamboo Dome
Professional Works
About Me
Urban Cognition
A Board Game with City Elements in Paris[Date]
2022.7 - 2022.9
[Location]
Paris, Franch
[Team Member]
Yueqi Hu, Xuejing Zhou
[Keywords]
Image Classification, Urban Data Analysis, Augmented Reality, Board Game Design
The complex urban historical process of Paris has led to difficulties in recognizing its urban plan and architecture style. The project aims to design a board game using machine learning to classify and filter both plans and facades, according to land use and historical features, for a better data-driven informatic cognition of cities. First, the texture of the Champs-Elysées plan was classified into public, residential, traffic, and green areas. Street view images, including facade information, were crawled from Google Map, trained, and predicted using CNN classification model depending on their historical styles. Second, we represent a puzzle game using the classified plan texture pieces as bottom, and the facade model as top modules. Finally, an urban cognition board game was designed according to the plan and facade image classification results, including plan texture puzzles, architecture modules, and AR view, to recall people's memory in the urban space.
Historical Layers in Urban Texture
1. Can urban texture be classified according to land use?
2. Can architectural facades be classified according to historical Style?
3. Can a game be designed to combine urban texture and architectural facades to enhance people’s cognition?
Methodology
1. Research scope: The Champs-Elysées in Paris, with over two hundred years of building history, has complex historical layers and city elements.
2. Data source: 500 pieces of urban texture were scrapped and divided randomly from Paris for training texture dataset, and 128 pieces in the research scope were divided to be predicted; 1689 pieces of architectural style were selected from arcDataset (Sun et al., 2022) for training facade dataset according to Paris history, and 12178 pieces of facade images were scrapped from Google Street View to be predicted. Public buildings’ points of interest (POI) from Open Street View were mapped as a reference, to select the representative buildings from each facades classification.
3. Machine Learning: Urban textures were classified into 5 classes with architecture facades into 11 classes by Resnet algorithm, to be further selected.
4. Board game design: The selected texture plans as the bottom puzzle, and the classified facades as the top building modules, finally formed a board game for recalling historical changes in urban space for citizens.
5. User test: 5 people were invited to complete preliminary user tests to verify the game’s effectiveness in helping people cognize the city.
5 classes including high-density residences, low-density residences, traffic, green and public area
11 classes including 11 classical architectural styles of Paris
Results 1. Urban texture can be classified according to land use into five types: low, high, traffic, green and public area, with public images being selected as the puzzle bottom.
2. Facades can be classified according to the architectural style exists in Paris into 11 types, in which the most public buildings from POI’s information were selected as the top modules with available AR view.
3. Our designed game, according to our pretest of 5 participants, can improve urban cognition for people to recall over 40% of the street imagery than before playing.
The 30 classified public pieces were selected as the puzzle bottom of our game.
The 20 most recognizable buildings in each class were selected as the top modules of our game, based
on public buildings’ POI.
The blanks and tabs of the puzzle are not meaningless and random, they are set as specific city memory points, including public buildings and grand plazas shared by the blocks on both sides of the city, helping the players to complete the layer 02.
/ Cognitive Flow