/Car-Body-Damage-Detection

Primary LanguageJupyter Notebook

Car Body Damage Detection : Detect Exterior Car Damage Using YOLOv5

Open In Colab

This project detects damage on the car body using the state-of-the-art YOLOv5 algorithm.

Project Steps

  1. Dataset of 100 images is collected.
  2. VOTT is used to add images annotations.
  3. Labels .txt files are generated.
  4. Dataset is splitted 9:1 for train and test.
  5. Model YOLOv5x is trained on the dataset.

Repo Content

  1. Custom Labeled Dataset
  2. Train Notebook Colab Link
  3. YOLOv5 .yaml File
  4. Results
  5. Annotations Guide

Train YOLOv5 on Custom Data

You can train YOLOv5 on your custom images through these steps:

  1. Place all the dataset images in one folder.
  2. Download VOTT.
  3. Follow the Annotations README.md.
  4. Set the dataset structure in Dataset README.md.
  5. Upload file to either Github or Google Drive.
  6. Clone the file to the Notebook.
  7. Run "CarBodyDamageDetection-YOLOv5.ipynb".
  8. Best model will be avilable at "/content/yolov5/runs/train/exp(i)/weights/best.pt".