Practical Deep Learning for Cloud, Mobile, and Edge

Chapter 14. Building the Purrfect Cat Locator App with TensorFlow Object Detection API

Bob gets frequent visits from a neighborhood feral cat. Those visits result in less than pleasant outcomes. You see, Bob has a nice fairly large garden that he tends to with great care. However, this furry little fella visits his garden every night and starts chewing on a bunch of plants. Months of hard work gets destroyed overnight. Clearly unhappy with this situation, Bob is keen on doing something about it.

Channeling his inner Ace Ventura (Pet Detective), he tries to stay awake at night to drive the cat away, but clearly, that’s not sustainable in the long term. After all, Red Bull has its limits. Reaching a breaking point, he decides to use the nuclear option: use AI in conjunction with his sprinkler system to literally “turn the hose” on the cat.

In his sprawling garden, he sets up a camera that can track the motion of the cat and turn on the nearest set of sprinklers to scare the cat away. He has an old phone lying around, and all he needs is a way to determine the cat’s location in real time, so he can control accordingly which sprinkler to trigger.

Hopefully, after a few unwelcome baths, as demonstrated in Figure 14-1, the cat would be disincentivized from raiding Bob’s garden.

Figure 14-1. Building an AI cat sprinkler system, like this Havahart Spray Away Motion Detector

Keen observers ...