Enterprise AI Analysis
Integrating Objective Segmentation and Subjective Perception to Predict Urban Landscape Preference: An XAI-Driven Approach
Abstract: Traditional urban landscape evaluations have primarily relied on either objective spatial metrics, such as the Green View Index (GVI), or subjective human surveys, often failing to capture the complex mechanisms of human environmental perception. This study proposes a novel Explainable Artificial Intelligence (XAI) framework that integrates objective physical configuration with subjective cognitive assessment to predict human landscape preference. Utilizing 159 urban landscape images, we extracted physical features via semantic segmentation (SegFormer) and psychological perceptions via a zero-shot vision-language model (CLIP). Our hybrid Random Forest model successfully bridged these dimensions, achieving moderate yet promising predictive performance (Rsquare = 0.442). SHAP (Shapley Additive exPlanations) analysis revealed that psychological perceptions—specifically Safety (0.104), Fascination (0.096), and Tranquility (0.080)—outperformed traditional objective metrics like GVI (0.067) in determining overall preference, while sub-model interpretation linked these psychological responses to specific physical elements such as buildings, sky openness, low vegetation, and water bodies. The findings suggest that urban green space design should move beyond maximizing greenery quantity and instead prioritize spatial compositions that induce psychological security, visual interest, and restoration. The proposed framework offers a scalable and interpretable tool for human-centered landscape assessment, while acknowledging limitations related to sample size, cultural generalizability, pretrained model bias, and reliance on static two-dimensional imagery.
0.442
Hybrid Model R² Score
0.104
Top Perceptual Driver (Safety SHAP)
0.067
Top Objective Driver (GVI SHAP)
SegFormer,
CLIP,
Random Forest Key AI Models Utilized
CLIP,
Random Forest Key AI Models Utilized
Executive Impact & Strategic Advantages
This research presents a groundbreaking XAI framework for understanding urban landscape preference, offering significant strategic advantages for urban planning, real estate development, and public policy.
Key Enterprise Takeaways
- Predictive Power: A novel XAI framework integrates objective physical data (SegFormer) with subjective psychological perceptions (CLIP) to accurately predict human landscape preference (Hybrid RF R²=0.442).
- Human-Centric Design: Revealed that psychological factors like safety, fascination, and tranquility are stronger predictors of preference than traditional metrics like Green View Index.
- Interpretable Insights: SHAP analysis provides clear, actionable insights, linking specific physical elements (e.g., buildings, water bodies, low vegetation, sky openness) to key psychological responses.
- Optimized Resource Allocation: Enables evidence-based urban green space planning, shifting focus from mere quantity of greenery to spatial compositions that enhance psychological well-being and restoration.
- Scalability & Efficiency: AI-driven methods offer a scalable and efficient alternative to labor-intensive traditional surveys for large-scale landscape assessment.
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Methodology
Key Findings
Design Implications
Limitations
Integrated XAI Framework
This study pioneers a dual-track XAI framework. It integrates SegFormer for objective physical feature extraction and CLIP for subjective psychological perception. These are combined using a hybrid machine learning model (Random Forest, XGBoost) to predict urban landscape preference with enhanced interpretability via SHAP analysis.
Enterprise Process Flow
Input & Data Collection
→
Feature Extraction (AI models)
→
Data Modeling
→
Sub-Model Analysis & Guidelines
Predictive Performance & Key Drivers
The hybrid model significantly improved predictive accuracy for landscape preference. SHAP analysis revealed that psychological perceptions are more influential than traditional objective metrics.
0.442
Achieved R² Score (Hybrid Random Forest Model)
| Model Configuration | Algorithm | R² Score |
|---|---|---|
| Objective Physical Metrics Only | Random Forest | 0.291 |
| Subjective Cognitive Perceptions Only | Multiple Linear Regression | 0.174 |
| Hybrid (Objective + Subjective) | Random Forest | 0.442 |
Psychological perceptions like Safety (0.104), Fascination (0.096), and Tranquility (0.080) consistently outranked the Green View Index (0.067) in overall preference prediction.
Evidence-Based Urban Design Guidelines
The SHAP sub-model analysis provides critical insights for designing urban green spaces that optimize human well-being, moving beyond mere greenery quantity.
Designing for Safety: Natural Surveillance & Openness
The presence of buildings (0.088, +) contributes positively to perceived safety, validating the "eyes on the street" principle. However, excessive building visibility negatively impacts overall preference. Optimal building proportion for safety is 5–15% of the visual field. Low vegetation (0.035, +) and sky openness (0.033, +) are also strong positive drivers for safety, contrasting with dense canopy trees that can impede sightlines.
Designing for Tranquility: Primacy of Blue-Green Infrastructure
Water bodies (0.138, +) are the dominant factor for tranquility, significantly more potent than GVI (0.036, +). Water features need to constitute at least 10-15% of the visual field to generate meaningful psychological benefit, indicating a threshold-gated effect. A GVI of 0.6-0.8 is identified as the optimum for tranquility and preference, with diminishing returns beyond 0.8.
Designing for Fascination: Dynamic Stimuli & Optimal Enclosure
Fascination is primarily driven by water bodies (0.013, +) and amenities (0.012, +), highlighting the need for dynamic visual stimuli and interactive elements. Highly enclosed scenes with low sky proportions (< 0.10) evoke strong fascination and refuge, resembling immersive forest-like environments. Intermediate sky openness (0.10-0.20) is penalized, suggesting a need for clear design intent towards either full enclosure or intentional openness (> 0.25-0.30).
Acknowledging Research Limitations
While robust, this study has several limitations:
- Limited Sample Size: 159 images, though subject to rigorous tuning, mean findings are moderate and exploratory.
- Source Heterogeneity: Pooled images from 16 studies introduce potential biases in image quality, cultural context, and participant demographics.
- CLIP Proxy Indicators: AI-derived cognitive scores are indirect proxies for psychological states and may not fully capture nuanced human perceptions, potentially inheriting Western aesthetic biases.
- Static Imagery: Reliance on two-dimensional images may not fully represent dynamic, multi-sensory environmental experiences.
Future research should address these by using larger, standardized datasets, validating AI scores with local surveys or physiological measures, and exploring multi-sensory data.
Calculate Your Potential AI ROI
Estimate the significant efficiency gains and cost savings your enterprise could achieve by integrating our XAI-driven landscape assessment.
Annual Cost Savings
$-
Annual Hours Reclaimed
-
Your AI Implementation Roadmap
A typical phased approach to integrating XAI-driven landscape assessment into your operations for maximum impact and minimal disruption.
Phase 01: Strategic Consultation & Needs Assessment
Collaborate to define specific objectives, identify key urban landscape contexts, and align the XAI framework with your existing planning and design workflows.
Phase 02: Data Integration & Custom Model Training
Integrate your proprietary landscape image data and fine-tune SegFormer and CLIP models to capture unique regional characteristics and aesthetic preferences.
Phase 03: XAI Framework Deployment & Initial Pilot
Deploy the integrated hybrid model and conduct a pilot assessment on a selected urban area, generating initial preference predictions and SHAP-driven design insights.
Phase 04: Performance Validation & Scalable Integration
Validate model performance against human expert evaluations, refine guidelines, and scale the framework across larger urban landscapes or multiple projects.
Ready to Transform Your Urban Planning?
Leverage the power of Explainable AI to create more human-centric, aesthetically pleasing, and psychologically restorative urban landscapes. Book a consultation with our experts today.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat