repository
Obstacle Detection on Roads based on Deep Learning
| DC Field | Value | Language |
|---|---|---|
| dc.contributor.advisor | 신동욱 | - |
| dc.contributor.author | 김상호 | - |
| dc.date.accessioned | 2025-01-25T01:35:54Z | - |
| dc.date.available | 2025-01-25T01:35:54Z | - |
| dc.date.issued | 2023-08 | - |
| dc.identifier.other | 32898 | - |
| dc.identifier.uri | https://dspace.ajou.ac.kr/handle/2018.oak/24367 | - |
| dc.description | 학위논문(석사)--수학과,2023. 8 | - |
| dc.description.tableofcontents | 1 Introduction 1 <br>2 Related Works 2 <br> 2.1 YOLO 3 <br> 2.2 Loss function for Bounding Box Regression 3 <br> 2.3 Loss function for Classification 4 <br>3 Methods 5 <br> 3.1 Class Imbalance 5 <br> 3.2 YOLOv7 5 <br> 3.3 Loss function 8 <br>4 Experiments 12 <br> 4.1 Dataset and Evaluation Metrics 13 <br> 4.2 Training Details 13 <br> 4.3 Experimental Results 15 <br> 4.4 Limitations 19 <br>5 Conclusion 21 | - |
| dc.language.iso | eng | - |
| dc.publisher | The Graduate School, Ajou University | - |
| dc.rights | 아주대학교 논문은 저작권에 의해 보호받습니다. | - |
| dc.title | Obstacle Detection on Roads based on Deep Learning | - |
| dc.type | Thesis | - |
| dc.contributor.affiliation | 아주대학교 대학원 | - |
| dc.contributor.alternativeName | Sangho Kim | - |
| dc.contributor.department | 일반대학원 수학과 | - |
| dc.date.awarded | 2023-08 | - |
| dc.description.degree | Master | - |
| dc.identifier.localId | T000000032898 | - |
| dc.identifier.url | https://dcoll.ajou.ac.kr/dcollection/common/orgView/000000032898 | - |
| dc.subject.keyword | Complete Intersection over Union | - |
| dc.subject.keyword | Median Frequency Balancing | - |
| dc.subject.keyword | Obstacle Detection | - |
| dc.subject.keyword | VariFocal Loss | - |
| dc.subject.keyword | YOLOv7 | - |
| dc.description.alternativeAbstract | In this paper, we implement a deep learning model to detect obstacles on roads for the blind individuals and autonomous delivery robots. We applied various methods to get model with high accuracy. Firstly, we used YOLOv7-tiny as a deep learning model, which demonstrated excellent performance due to its efficient architecture. Secondly, we used median frequency balancing to solve class imbalance in the dataset, resulting in an increase of 0.4 in mAP. We also conducted experiments with different bounding box regression losses, such as GIoU, DIoU, and CIoU, as well as classification losses, such as FL, QFL, and VFL, to improve the efficient training and performance of the model. As a result, CIoU, which considers all important factors in bounding box regression, and VFL, which effectively addresses foreground-background class imbalance, showed the best performance with 68.7 mAP, surpassing the baseline by over 3% in mAP. | - |
- Appears in Collections:
- Graduate School of Ajou University > Department of Mathematics > 3. Theses(Master)
- Files in This Item:
- There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.
-
- License
- STATISTICS
- Total Visit :5,244,000
- Total Download :2,137
- Today View :11,834
