TL;DR:
- Tennis Smith’s Cat Doorbell utilizes on-device machine learning (ML) to solve the challenges of cat-human communication.
- Smith’s cat would often get forgotten outside, leading to the creation of the Cat Doorbell.
- The device combines sound and sight cues to accurately detect when the cat wants to come inside.
- It uses a Python-powered program and TensorFlow examples for on-device ML to identify the cat.
- The Cat Doorbell sends a text message alert when the cat is identified and pauses to prevent repeated alerts.
- The project is open-source and available on GitHub under an Apache 2.0 license.
Main AI News:
Tennis Smith’s groundbreaking Cat Doorbell employs on-device machine learning (ML) to tackle the age-old issue of feline-human communication, all while adhering to a style befitting a business magazine. This ingenious device ensures that your cat’s needs are met promptly, eliminating the frustration of missed cues and false alarms.
The Challenge: A Vocal and Playful Feline Smith’s cat, like many others, enjoyed the freedom of the outdoors but had a penchant for causing confusion. “We have a cat that likes to go out on our enclosed patio,” Smith explains. “When he is ready to come in, he will stand next to the door and meow (yell). We open the door and let him in. No problem.” However, the cat’s unpredictable behavior created two major problems.
Problem #1: Forgetfulness The cat would often remain outside for extended periods, causing the family to forget about him. “More than once, he was outside yelling, and we were oblivious,” laments Smith. This forgetfulness needed addressing.
Problem #2: False Alarms The cat’s tendency to yell for the sheer enjoyment of it led to numerous “false alarms.” Smith remarks, “He will lay on the ground wallowing and yelling just for the fun of it.” These false alarms frustrated the family. It was clear that a solution was required to distinguish between genuine requests and playful antics.
The Ingenious Solution: The Cat Doorbell To address these challenges, Smith created the Cat Doorbell. The initial version of the device relied solely on sound cues, triggering alerts whenever the cat meowed. However, this approach was prone to false alarms. The solution was to incorporate a camera, ensuring that alerts were only sent when both sight and sound confirmed the cat’s presence. The magic ingredient: a Python-powered program that harnessed two TensorFlow example projects for on-device machine learning to identify the cat by both sight and sound.
The Cat Doorbell in Action Smith describes the device as “essentially a small state machine.” It passively listens for the sound of the cat meowing. When the distinctive sound is detected, it activates the camera. If the onboard light sensor detects darkness, an LED strip illuminates the area. For 45 seconds, the Cat Doorbell uses the camera to identify the cat. If no cat is identified, it reverts to passive listening.
When the cat is successfully identified during the 45-second window, a text message is sent to alert the owner. The system then pauses for two minutes to prevent repeated alerts. If darkness is detected, the LED light remains on until after the pause, after which the Cat Doorbell resumes passive listening.
Innovative and Accessible Smith’s comprehensive project write-up and the source code, available under an Apache 2.0 license on his GitHub repository, make this innovative Cat Doorbell accessible to other pet owners who seek a solution to similar challenges. Tennis Smith’s creation demonstrates the power of on-device machine learning to enhance our everyday lives, even in unexpected ways.
Conclusion:
Tennis Smith’s innovative Cat Doorbell, incorporating on-device machine learning, addresses the common challenges of cat owners. It introduces a novel solution to a niche market, showcasing the potential of ML in enhancing pet-human interactions. As the pet technology market continues to grow, such inventions may inspire similar developments and cater to the evolving needs of pet owners.
