The images that we will use to test the pretrained Inception-v3 deep CNN have been selected from the Open Images v4 dataset, a collection of over 9 million images that have been released under the Creative Common Attribution license, and which may be found at https://storage.googleapis.com/openimages/web/index.html.
In the GitHub repository accompanying this book, you can find 30 images of birds (image-recognition-data/birds) and 30 images of planes (image-recognition-data/planes) respectively. Figure 7.15 shows a couple of examples of the images that you might find in these test datasets: