Abstract:This study aims to compare the effectiveness of a robust ensemble model with the state-of-the-art ONE-PEACE Large Language Model (LLM) for accurate detection of sidewalks. Accurate sidewalk detection is crucial in improving road safety and urban planning. The study evaluated the model's performance on Cityscapes, Ade20k, and the Boston Dataset. The results showed that the ensemble model performed better than the individual models, achieving mean Intersection Over Union (mIOU) scores of 93.1\%, 90.3\%, and 90.6\% on these datasets under ideal conditions. Additionally, the ensemble model maintained a consistent level of performance even in challenging conditions such as Salt-and-Pepper and Speckle noise, with only a gradual decrease in efficiency observed. On the other hand, the ONE-PEACE LLM performed slightly better than the ensemble model in ideal scenarios but experienced a significant decline in performance under noisy conditions. These findings demonstrate the robustness and reliability of the ensemble model, making it a valuable asset for improving urban infrastructure related to road safety and curb space management. This study contributes positively to the broader context of urban health and mobility.

What problem does this paper attempt to address?

### Problems the Paper Attempts to Solve This paper aims to compare the effectiveness of a robust ensemble model with the state-of-the-art large language model (ONE-PEACE LLM) in sidewalk detection. Accurate sidewalk detection is crucial for improving road safety and urban planning. The study evaluates the model's performance on three datasets: Cityscapes, Ade20k, and the Boston Dataset. The results show that under ideal conditions, the ensemble model outperforms individual models, achieving mean Intersection over Union (mIOU) scores of 93.1%, 90.3%, and 90.6% on these datasets, respectively. Additionally, under challenging conditions such as salt-and-pepper noise and speckle noise, the ensemble model maintains consistent performance, showing only a gradual decline. In contrast, ONE-PEACE LLM slightly outperforms the ensemble model under ideal conditions but shows a significant performance drop under noisy conditions. These findings demonstrate the robustness and reliability of the ensemble model, making it a valuable asset for improving urban infrastructure related to road safety and curbside space management. The study has a positive impact on urban health and mobility. ### Specific Problem Description 1. **Objectives**: - Accurately detect sidewalks to enhance road safety and pedestrian safety, and facilitate the smooth operation of autonomous vehicles. - Manage curbside space by separating sidewalk and vehicle areas to reduce accident risks and minimize traffic congestion caused by improper parking. 2. **Background Technology**: - Classical computer vision methods such as affine transformation and dynamic contour models have achieved 80% accuracy in sidewalk detection but still have room for improvement. - Common object detection methods like YOLO, Faster-RCNN, and Single Shot Multi-box Detector struggle to accurately identify sidewalks in complex urban environments. 3. **Proposed Method**: - An image segmentation method based on ensemble learning is proposed, leveraging the strengths of multiple segmentation models to accurately detect various sidewalks in different urban environments. - This method combines Hierarchical Adaptive Mean Shift (HAMM), DeepLabV3, and YOLACT to create an ensemble model that excels in accuracy, precision, and noise resistance. 4. **Experimental Results**: - The ensemble model outperforms individual models on the Cityscapes, Ade20k, and Boston Dataset, particularly showing more stable performance under noisy conditions. - The robustness of the model under noisy conditions is validated through the ensemble learning approach, demonstrating its potential for real-world applications. ### Summary The main contribution of this paper is the proposal of a simple yet effective ensemble model that leverages the combination of advanced models to surpass the performance of individual models. It also introduces the large language model ONE-PEACE for comparative analysis for the first time. The study showcases the superior performance of the ensemble model under noisy conditions and provides suggestions for future research directions.

Precise and Robust Sidewalk Detection: Leveraging Ensemble Learning to Surpass LLM Limitations in Urban Environments

Research on lightweight pavement disease detection model based on YOLOv7

Revolutionizing Urban Safety Perception Assessments: Integrating Multimodal Large Language Models with Street View Images

Illumination Invariance Adaptive Sidewalk Detection Based on Unsupervised Feature Learning

ESCORT: Fine-Grained Urban Crime Risk Inference Leveraging Heterogeneous Open Data

CitySurfaces: City-scale semantic segmentation of sidewalk materials

Sidewalk Measurements from Satellite Images: Preliminary Findings

Pedestrian Detection with Spatially Pooled Features and Structured Ensemble Learning

An End-to-End Framework for Unsupervised Pose Estimation of Occluded Pedestrians

Automatic concrete sidewalk deficiency detection and mapping with deep learning

Robust multi-modal pedestrian detection using deep convolutional neural network with ensemble learning model

DELTA: Integrating Multimodal Sensing with Micromobility for Enhanced Sidewalk and Pedestrian Route Understanding

Leveraging 3D LiDAR Sensors to Enable Enhanced Urban Safety and Public Health: Pedestrian Monitoring and Abnormal Activity Detection

A New Urban Objects Detection Framework Using Weakly Annotated Sets

VLPD: Context-Aware Pedestrian Detection via Vision-Language Semantic Self-Supervision

Accurate detection of vehicle, pedestrian, cyclist and wheelchair from roadside light detection and ranging sensors

A low complexity contextual stacked ensemble-learning approach for pedestrian intent prediction

Self-Supervised Sidewalk Perception Using Fast Video Semantic Segmentation for Robotic Wheelchairs in Smart Mobility

Off The Beaten Sidewalk: Pedestrian Prediction In Shared Spaces For Autonomous Vehicles

Smart City Transportation: Deep Learning Ensemble Approach for Traffic Accident Detection

YOLO-ABD: A Multi-Scale Detection Model for Pedestrian Anomaly Behavior Detection