A deep learning-powered image classifier that identifies 10 animal species with 95.9% accuracy using transfer learning.

• Dual Input Methods: Upload images directly or provide image URLs
• Real-time Inference: Instant predictions with confidence scores
• Transparency: View probability distribution across all classes
• User-Friendly Interface: Clean, intuitive Streamlit interface

Butterfly • Cat • Chicken • Cow • Dog • Elephant • Horse • Sheep • Spider • Squirrel

Model Architecture: ResNet50 with transfer learning
Training Dataset: 26,179 images across 10 classes
Performance Metrics:

• Test Accuracy: 95.9%
• Training Time: <5 minutes (NVIDIA RTX 5070 Ti)

Approach: Fine-tuned a pre-trained ResNet50 model on a custom animal dataset, freezing early layers and training only the final classification layer for efficient learning.

• Framework: PyTorch 2.11
• Interface: Streamlit
• Language: Python 3.11
• Deployment: Hugging Face Spaces

1. Choose input method: file upload or image URL
2. Provide an animal image
3. View prediction with confidence score and full probability breakdown

Project Type: Computer Vision • Deep Learning • Transfer Learning
Status: Production-ready deployment

Built as part of my machine learning portfolio to demonstrate end-to-end ML project capabilities.