This project focuses on detecting and classifying car models using the Stanford Car Dataset. The implementation is done in a Jupyter Notebook using transfer learning with ResNet18 to perform fine-grained classification of car images.
car-model-detection/
├── car-model-detection.ipynb # Main notebook with the complete pipeline
└── README.md- Image preprocessing and data augmentation
- Transfer learning using ResNet18
- Training and validation on a labeled dataset of car models
- Visualization of accuracy and loss during training
- Evaluation on test images using confusion matrix and classification report
Install the following Python libraries:
pip install numpy pandas matplotlib seaborn opencv-python scikit-learn torch torchvision-
Training accuracy: ~98.74%
-
Validation accuracy: ~88.36%%
Confusion matrix for the first 20 classes:
Sample predictions on unseen test images:
Contributions are welcome! Please feel free to open issues or pull requests.
Contact: arashkeshavarzx@gmail.com