| Title |
Computer Vision-Based Object Detection and Depth Estimation to Generate Point Clouds and Estimate Distances |
| Authors |
엄태현(Tae-Hyun Eom) ; 백우진(Woojin Paik) |
| DOI |
https://doi.org/10.5370/KIEE.2025.74.1.164 |
| Keywords |
Distance Estimation; Object Detection; Depth Estimation; Point Cloud; Stereo Camera; Mono Camera |
| Abstract |
This research proposes a distance estimation method using Mono Camera-based object detection and depth estimation to generate Point Cloud data. The study aims to enhance the applicability of Mono Cameras in autonomous vehicles and robots, reducing costs compared to Stereo Camera systems. The method utilizes YOLOv8 for object detection and Depth Anything V2 for depth estimation. Results indicate that while the proposed method offers potential, it exhibits higher error rates in distance estimation compared to Stereo Camera-based approaches, primarily due to the limitations of current depth estimation technologies. The study highlights the need for further improvements in depth estimation models, particularly to address environmental factors such as lighting. Additionally, the research demonstrates that combining depth estimation with bounding box methods helps reduce estimation errors, showing promise for more stable performance. Future work will focus on improving depth estimation accuracy and making the proposed method more efficient for real-time applications, with the potential to integrate into Visual SLAM, a key technology in autonomous driving systems. |