In this notebook an algorithm is developed to classify images of humans and dogs, that could be used as part of a mobile or web app. The code accepts any user-supplied image as input. If a dog is detected in the image, it provides an estimate of the dog's breed. If a human is detected, it provided an estimate of the dog breed that is most resembling. The image below displays a sample output.
The following datasets are used to train the models:
-
Dog dataset consists of 8351 dog images.
-
Human dataset consists of 13233 human images.
The notebook is split into the following steps:
- Step 0: Import Datasets
- Step 1: Detect Humans
- Step 2: Detect Dogs
- Step 3: Create a CNN to Classify Dog Breeds (from Scratch)
- Step 4: Create a CNN to Classify Dog Breeds (using Transfer Learning)
- Step 5: Create the Algorithm
- Step 6: Test the Algorithm