Istanbul: A Way Out




"A Way Out" is a research-driven architectural project that explores the post-earthquake emergency conditions of Istanbul. Rather than focusing solely on building collapses, the project examines whether streets remain accessible for evacuation and aid distribution.Istanbul, a city of 20 million, has developed chaotically, with dense informal settlements and unpredictable urban patterns.

Official hazard assessments highlight structural risks, but the 2023 Turkey earthquake demonstrated that blocked streets can be just as deadly—preventing rescue teams from reaching those in need.To address this issue, urban observation and AI-based analysis were combined. Streets were studied through object detection and semantic segmentation models, identifying urban elements that could obstruct emergency access—such as narrow roads, traditional architectural features (cumbas), street furniture, parked cars, and pavement usage. These elements were categorized and scored based on their likelihood of contributing to post-disaster inaccessibility.

The outcome is a comparative risk mapping system, highlighting streets where obstructions are more likely to occur. Rather than predicting collapses, this tool provides a new layer of analysis for emergency planning, assisting authorities in identifying alternative escape routes and critical urban bottlenecks.

"A Way Out" serves as an invitation to rethink urban resilience—not just in terms of structural stability but through the lens of accessibility, emergency response, and post-disaster survival.


The Team 
We are a multinational group of architects, all under the age of 30, brought together by a shared commitment to design innovation and critical discourse. Our members include PhD candidates, researchers, and practicing architects from diverse academic and professional backgrounds. United by the interest in architectural competitions, seminars, and small-scale research initiatives, the team regularly engages in collaborative discussions to explore pressing topics within the built environment. This project emerged from one such dialogue—sparked by reflections on seismic resilience, particularly in light of the devastating 2023 earthquake in Turkey.





Point 0





















The Problem: A Hidden Crisis





















Our Approach: Seeing Beyond the Obvious
The 2023 earthquake in Hatay intensified widespread concern about the long-anticipated major earthquake projected to strike Istanbul—a city of nearly 20 million. Unlike many European cities, Istanbul has developed in a largely unplanned and fragmented manner, resulting in significant vulnerabilities. Current data indicates that around 2 million residents are at direct risk. It was this alarming context that prompted the initiation of our research, driven by the urgent need to better understand and address the city's spatial and infrastructural fragilities.




Earthquake hazard assessments have traditionally focused on the potential for building collapse. However, the events of 2023 demonstrated that the most severe consequences often arise after the shaking stops—when poorly coordinated emergency responses delay the delivery of critical aid.

What happens if a city lacks clearly defined escape routes? If streets are blocked by elements other than collapsed buildings? In dense, informally developed urban areas with challenging terrain, such conditions can pose serious, even life-threatening risks.




A different approach to analyzing Istanbul was pursued — one that shifted the focus away from structural failure. Instead, the investigation centered on questions such as: Could a street become impassable, even if buildings remain intact? Might parked vehicles, street furniture, or narrow passageways obstruct access for emergency response? And could such conditions be systematically mapped?

To address these questions, artificial intelligence was introduced as a means to rapidly interpret the spatial and behavioral patterns embedded within the urban fabric.

Study areas were selected based on research identifying correlations between earthquake risk and socioeconomic vulnerability, with particular attention given to neighborhoods characterized by high density, informal urban development, and irregular street networks.




Credits Eren Sezer
MSc Architect / Researcher
Turkey
eren.sezer@hotmail.com


Egemen Sezer
MSc Architect / Researcher
Turkey
egemensezeruni@gmail.com


Nour Fneich
MSc Architect / Researcher
Lebanon
nourfneish1@gmail.com


Raşit Eren Cangür
Phd Architect / Researcher
Turkey
rasiterencangur@gmail.com


Andrei Calin Teodorescu
MSc Architect / Researcher
Romania
andreiteodorescu1997@gmail.com


Sonya Ragimova
MSc Architect / Researcher
Russia
arch.sona.ragimova@gmail.com


Nicolo Carlini
MSc Architect / Researcher
Italy
carlininicolo@gmail.com

Exhibitions
La Biennale Architettura di Venezia 2025
Istanbul A Way Out
May - September 2025











Last Updated 24.10.31
48d 7h 37m 58s since last earthquake







We want to show a dense urban block where everyday objects—like parked cars, trees, kiosks, and architectural extensions—begin to form potential barriers in an artistic method. What might seem like harmless urban details take on a new meaning when access becomes a matter of life and death. In post-earthquake scenarios, these accumulations can block aid routes or delay evacuations. By mapping these elements visually, the project highlights how the problem of accessibility is often rooted in daily urban clutter—not just in collapsed structures.



2.Istanbul
 Istanbul is a city more populous than any European capital, layered with distinct socio-economic zones. It has continuously expanded throughout history, shaped by waves of migration.To train this model we choose 7 Case Studies Areas of Istanbul to improve variety of objects.After collecting enough data,We startedd to search for solid source for liability, thanks to our team members who have lived in Istanbul for many years, we hold a deep understanding of its urban structure and varying levels of livability.

One question guided our analysis: Do the results of our AI model align with the widely known socio-economic disparities between Istanbul's neighborhoods?

To explore this, we selected three contrasting areas:

  • Dolapdere is a densely populated, socio-economically disadvantaged area with limited urban planning and infrastructure.

  • Ortaköy represents a historically rich neighborhood with preserved architectural texture and a relatively strong socio-economic profile, yet with moderate urban development.

  • Göztepe is a newly planned district with a low population density and a well-structured urban layout, reflecting a high socio-economic standard.

By incorporating this socio-spatial context into our training process, we aimed to see whether the risk factors identified by the AI could reflect more than just physical obstructions—and potentially reveal how urban inequality manifests in emergency accessibility.





3.YOLO Model
In this Illustruation/Google Maps Image, YOLOv model have identified elements that could block emergency access—like bay windows, narrow pavements, parked cars, or even lamp posts. After training the AI model with Google Images, Object Detection starts to see the invisible barriers that might slow down help when every second matters.





4. Street Section Diagram

 One of the visual analysis tools developed as part of the project is this street section diagram, which illustrates potential risk elements identified through AI-based models. Parked cars, trees, bus stops, narrow sidewalks, and architectural protrusions such as bay windows (cumbas) are highlighted using bounding boxes. The horizontal axis, labeled “Density Interaction Factor,” represents the intensity of spatial interaction among these elements, while the vertical axis, “Risk Factor Weight,” indicates each element’s potential to obstruct emergency accessibility.

This visual representation directly aligns with the core approach of A Way Out—a project that reconsiders urban emergency preparedness through spatial accessibility rather than solely structural collapse. By mapping these obstruction risks, the project offers a multilayered assessment system that helps authorities identify critical bottlenecks and alternative escape routes in the event of a disaster.






5.Street Graph
 This graph illustrates the relationship between the physical width of a street and its associated risk score. Narrow and densely occupied streets correspond to higher risk levels, while wider, more open roads are considered safer in terms of emergency accessibility.

This finding reinforces the core argument of A Way Out: in disaster scenarios, the accessibility of streets can be just as critical as the structural integrity of buildings. The project emphasizes that post-earthquake survival depends not only on what collapses, but also on whether rescue teams can reach those in need.





6. Total Risk Map
 These are risk analysis maps for the neighborhoods of Dolapdere, Ortaköy, and Göztepe. Each area was evaluated using grid systems of varying sizes, and risk levels were calculated using a “Total Risk Score” formula. At the bottom of each map, the formula is detailed: the risk score is normalized based on the proportion of obstructed street, sidewalk, and open space areas within the total image area.

This methodology enables a comparative understanding of neighborhood-level accessibility vulnerabilities, highlighting how spatial configurations impact emergency response potential. By applying this scoring system across different urban textures, A Way Out aims to create a scalable and data-driven tool for post-disaster risk assessment and planning.