Skip to content

Advanced Deep Learning model using ResNet50 for accurate classification of dog and cat breeds from the Oxford-IIIT Pet Dataset. Showcases state-of-the-art techniques in computer vision and provides insights into model optimization and performance evaluation.

License

Notifications You must be signed in to change notification settings

kiansahafi/PetBreedClassifier-ResNet50

Repository files navigation

Oxford-IIIT Pet Dataset Classification Challenge 🐾

Overview

a computer vision project developed for classifying dog and cat breeds using the Oxford-IIIT Pet Dataset. 🐶🐱

Data Preparation📊

Dataset: 37 dog/cat breeds with 200 images each.
Augmentation: Increased robustness through data augmentation, creating a richer dataset for model training.🔄

Model Development🛠️

Initial Attempts: Custom model architectures; trialed VGG-16 and VGG-19.
Final Choice: ResNet50-V2, which significantly improved accuracy.🚀
Architecture: ResNet50-V2 (pre-trained, partially frozen), followed by fully connected layers, dropout, and a final output layer.

Training and Evaluation📈

Validation Accuracy: Achieved a high accuracy (specific percentage detailed in the project).✅
Optimization: Used ADAM optimizer for better performance.⚙️
Learning Rate Scheduler: Implemented to optimize training process.📉

Results📋

Tip

Validation Accuracy: 87%
Validation Loss: 41%

Keeshond Birman
alt text alt text

Confusion Matrix: Indicates good performance, with some confusion between similar breeds.🧩
Sample Outputs: Showcases the model's breed classification capabilities.👀

Improvements🌟

Data Distribution: Increasing the representation of confused breeds could enhance accuracy. However, dataset limitations restrict this approach.📚

Shows an illustrated sun in light mode and a moon with stars in dark mode.

Further Tuning: Optimizing learning rate and further experimenting with model architecture and training strategies.🔧

How to Use🖥️

  1. Clone the repo.
  2. Install dependencies (list dependencies here).
  3. Run the model (provide command or script).

Contributing🤝

Feel free to contribute! Open an issue or submit PRs.

License📝

MIT License

About

Advanced Deep Learning model using ResNet50 for accurate classification of dog and cat breeds from the Oxford-IIIT Pet Dataset. Showcases state-of-the-art techniques in computer vision and provides insights into model optimization and performance evaluation.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published