Detect the breed of a dog provided the image of a dog. If the image is of a human, identify the dog breed which resembles the person.

Dog Detection is a popular starter Computer Vision problem. This project aims to use some of the most common methods for Dog Detection in images.

• OpenCV for Python
• Keras
• Scikit-Learn
• Matplotlib
• Numpy

• haarcascades
  -- haarcascade_frontalface_alt.xml : Features for HaarCascadeClassifier
• images
  -- h1.jpg
  -- h2.jpg
  -- h3.jpg
  -- d1.jpg
  -- d2.jpg
  -- d3.jpg
• saved_models
  -- weights.best.resnet50.hdf5 : Saved Weights for ResNet50 model
• dog_app.ipynb: IPython Notebook containing code for Dog Detection
• dog_app.html: IPython Notebook saved as .html
• extract_bottleneck_features.py: Script to extract bottleneck features for transfer learning

1. The model achieves an accuracy of 80.7416% on the test set.
2. The model is able to differentiate between the image of a dog, a human and any other image.
3. The model outputs the dog breed which resembles the provided human image.
4. A Medium blog post on this project was published here: https://medium.com/@sharmadivyanshu1709/dog-breed-classifier-10-minute-challenge-6a78413ea2a2?source=friends_link&sk=fe44b530d48c6187f416070034d059e5

Credit to Udacity for providing the starter code.