13 research outputs found
Songdo Traffic: High Accuracy Georeferenced Vehicle Trajectories from a Large-Scale Study in a Smart City
No full textThe Songdo Traffic dataset delivers precisely georeferenced vehicle trajectories captured through high-altitude bird's-eye view (BeV) drone footage over Songdo International Business District, South Korea. Comprising approximately 700,000 unique trajectories, this resource represents one of the most extensive aerial traffic datasets publicly available, distinguishing itself through exceptional temporal resolution that captures vehicle movements at 29.97 points per second, enabling unprecedented granularity for advanced urban mobility analysis.
The dataset consists of four primary components:
- Trajectory Data: 80 ZIP archives containing high-resolution vehicle trajectories with georeferenced positions, speeds and acceleration profiles, and other metadata.
- Orthophoto Cut-Outs: High-resolution (8000×8000 pixel) orthophoto images for each monitored intersection, used for georeferencing and visualization.
- Road and Lane Segmentations: CSV files defining lane polygons within road sections, facilitating mapping of vehicle positions to road segments and lanes.
- Sample Videos: A selection of 4K UHD drone video samples capturing intersection footage during the experiment.
The dataset was collected as part of a collaborative multi-drone experiment conducted by KAIST and EPFL in Songdo, South Korea, from October 4–7, 2022.
- A fleet of 10 drones monitored 20 busy intersections, executing advanced flight plans to optimize coverage.
- 4K (3840×2160) RGB video footage was recorded at 29.97 FPS from altitudes of 140–150 meters.
- Each drone flew 10 sessions per day, covering peak morning and afternoon periods.
- The experiment resulted in 12TB of 4K raw video data.
More details on the experimental setup and data processing pipeline are available in the published article:
Robert Fonod, Haechan Cho, Hwasoo Yeo, Nikolas Geroliminis (2025). Advanced computer vision for extracting georeferenced vehicle trajectories from drone imagery, Transportation Research Part C: Emerging Technologies, vol. 178, 105205. DOI: 10.1016/j.trc.2025.105205LUTS
Geo-trax: A Comprehensive Framework for Georeferenced Vehicle Trajectory Extraction from Drone Imagery
No full textGeo-trax (GEO-referenced TRAjectory eXtraction) is a comprehensive pipeline for extracting high-accuracy georeferenced vehicle trajectories from high-altitude drone imagery. Designed specifically for quasi-stationary aerial monitoring in urban traffic scenarios, Geo-trax transforms raw, bird’s-eye view video footage into precise, real-world vehicle trajectories. The framework integrates state-of-the-art computer vision and deep learning modules for vehicle detection, tracking, and trajectory stabilization, followed by a georeferencing stage that employs image registration to align the stabilized video frames with an orthophoto. This registration enables the accurate mapping of vehicle trajectories to real-world coordinates. The resulting pipeline supports large-scale traffic studies by delivering spatially and temporally consistent trajectory data suitable for traffic behavior analysis and simulation. Geo-trax is optimized for urban intersections and arterial corridors, where high-fidelity vehicle-level insights are essential for intelligent transportation systems (ITS) and digital twin applications.LUTS5.
Can Urban Air Mobility Become a Public Transportation Option?: Understanding Sociodemographic Characteristics and Urban Air Mobility Adoption Readiness in Seoul, South Korea
No full textSongdo Vision: Vehicle Annotations from High-Altitude BeV Drone Imagery in a Smart City
No full textThe Songdo Vision dataset provides high-resolution (4K, 3840×2160 pixels) RGB images annotated with categorized axis-aligned bounding boxes (BBs) for vehicle detection from a high-altitude bird’s-eye view (BeV) perspective. Captured over Songdo International Business District, South Korea, this dataset consists of 5,419 annotated video frames, featuring approximately 300,000 vehicle instances categorized into four classes:
- Car (including vans and light-duty vehicles)
- Bus
- Truck
- Motorcycle
This dataset can serve as a benchmark for aerial vehicle detection, supporting research and real-world applications in intelligent transportation systems, traffic monitoring, and aerial vision-based mobility analytics. It was developed in the context of a multi-drone experiment aimed at enhancing geo-referenced vehicle trajectory extraction.LUTS
Generalizable Novel-View Synthesis using a Stereo Camera
No full textIn this paper, we propose the first generalizable view synthesis approach
that specifically targets multi-view stereo-camera images. Since recent stereo
matching has demonstrated accurate geometry prediction, we introduce stereo
matching into novel-view synthesis for high-quality geometry reconstruction. To
this end, this paper proposes a novel framework, dubbed StereoNeRF, which
integrates stereo matching into a NeRF-based generalizable view synthesis
approach. StereoNeRF is equipped with three key components to effectively
exploit stereo matching in novel-view synthesis: a stereo feature extractor, a
depth-guided plane-sweeping, and a stereo depth loss. Moreover, we propose the
StereoNVS dataset, the first multi-view dataset of stereo-camera images,
encompassing a wide variety of both real and synthetic scenes. Our experimental
results demonstrate that StereoNeRF surpasses previous approaches in
generalizable view synthesis.Comment: Accepted to CVPR 2024. Project page URL:
https://jinwonjoon.github.io/stereonerf
Advanced computer vision for extracting georeferenced vehicle trajectories from drone imagery
No full textThis paper presents a comprehensive framework for extracting georeferenced vehicle trajectories from high-altitude drone imagery, addressing key challenges in urban traffic monitoring and the limitations of traditional ground-based systems. Our approach integrates several novel contributions, including a tailored object detector optimized for high-altitude bird’s-eye view perspectives, a unique track stabilization method that uses detected vehicle bounding boxes as exclusion masks during image registration, and an orthophoto and master frame-based georeferencing strategy that enhances consistent alignment across multiple drone viewpoints. Additionally, our framework features robust vehicle dimension estimation and detailed road segmentation, enabling comprehensive traffic dynamics analysis. Conducted in the Songdo International Business District, South Korea, the study utilized a multi-drone experiment covering 20 intersections, capturing approximately 12TB of ultra-high-definition video data over four days. The framework produced two high-quality datasets: the Songdo Traffic dataset, comprising approximately 700,000 unique vehicle trajectories, and the Songdo Vision dataset, containing over 5000 human-annotated images with about 300,000 vehicle instances categorized into four classes. Comparisons with high-precision sensor data from an instrumented probe vehicle highlight the accuracy and consistency of our extraction pipeline in dense urban environments. The public release of the Songdo Traffic and Songdo Vision datasets, along with the complete source code for the extraction pipeline, establishes new benchmarks in data quality, reproducibility, and scalability in traffic research. The results demonstrate the potential of integrating drone technology with advanced computer vision methods for precise and cost-effective urban traffic monitoring, providing valuable resources for developing intelligent transportation systems and enhancing traffic management strategies.LUT
NAVIBox: Real-Time Vehicle–Pedestrian Risk Prediction System in an Edge Vision Environment
No full textThis study introduces a novel system, termed NAVIBox, designed to proactively identify vehicle–pedestrian risks using vision sensors deployed within edge computing devices in the field. NAVIBox consolidates all operational components into a single unit, resembling an intelligent CCTV system, and is built upon four core pipelines: motioned-video capture, object detection and tracking, trajectory refinement, and predictive risk recognition and warning decision. The operation begins with the capture of motioned video through a frame difference approach. Road users are subsequently detected, and their trajectories are determined using a deep learning-based lightweight object detection model, in conjunction with the Centroid tracker. In the trajectory refinement stage, the system converts the perspective of the original image into a top view and conducts grid segmentation to capture road users’ behaviors precisely. Lastly, vehicle–pedestrian risks are predicted by analyzing these extracted behaviors, and alert signals are promptly dispatched to drivers and pedestrians when risks are anticipated. The feasibility and practicality of the proposed system have been verified through implementation and testing in real-world test sites within Sejong City, South Korea. This systematic approach presents a comprehensive solution to proactively identify and address vehicle–pedestrian risks, enhancing safety and efficiency in urban environments
High-Precision Trajectory Extraction Using Advanced Computer Vision From a Multi-Drone Experiment in South Korea
No full textIn this study, we develop a complete framework for accurate trajectory extraction for multi-drone experiments with an application in the Songdo International Business District, South Korea. The raw bird’s eye view drone footage was processed and segmented, resulting in a substantial database of 4K videos. We annotate a total of 5,419 video frames to create the exhaustive Songdo visual dataset, a valuable resource for object detection validation. Our methodology involves a transformation of raw drone footage into comprehensive traffic data via a robust vehicle trajectory extraction pipeline. Throughout the experiment, meticulous procedures were employed to ensure data integrity and precision. Key findings from the study underline the potential of drone technology combined with computer vision tools for reliable and detailed traffic monitoring in smart cities. Notably, we leveraged the high-precision autonomous vehicle trajectories from Stanford to quantitatively validate our pipeline, further reinforcing the robustness of our results. As part of our commitment to open science, we plan to release both the Songdo visual and trajectory datasets to the public.LUTSPaper Number: TRBAM-24-0427
