Vehicle Damage Detection Using Deep Learning With Yolo Algorithm |
Author(s): |
| Dhrupad Thanvi , KJ Somaiya College Of Engineering; Soham Loke, KJ Somaiya College Of Engineering; Hitesh Bhanushali, KJ Somaiya College Of Engineering; Yash Musale, KJ Somaiya College Of Engineering |
Keywords: |
| Vehicle Damage Detection, Deep Learning, YOLO Algorithm |
Abstract |
|
1) Data Preparation: Data Quality Having good quality annotated photographs is essential. Include a variety of car models, perspectives, and damage types (scratches, dents, broken parts, etc.). 2) Diversity: The dataset should represent a variety of backdrops, climates, and lighting conditions in order to improve model generalization. Tools for Annotation: Applications such as LabelImg, Roboflow, or CVAT can be used to expedite the annotation process. Class Imbalance: Address class imbalance (e.g., more minor scratches vs fewer damaged components) to prevent bias in forecasts. 3) YOLO versions 7 and 8 Features: YOLOv7: Very quick and accurate. emphasizes extremely precise real-time detection, which qualifies it for applications such as insurance and on-site inspection. YOLOv8: More user-friendly and with improved inference and training support. improved model. |
Other Details |
|
Paper ID: IJSRDV12I120011 Published in: Volume : 12, Issue : 12 Publication Date: 01/03/2025 Page(s): 10-13 |
Article Preview |
|
|
|
|
