Assess your aquarium’s well being with an AI-enabled ultrasonic sensor

Beneath the floor of any physique of water, dangerous quantities of poisonous gases and contaminates can accumulate, which ends up in a loss in fish and plant populations if not fastened shortly. However as a result of most water testing, particularly in aquariums, is finished totally on the floor, very important info will get missed. Kutluhan Aktar’s automated testing system goals to handle these considerations via its harnessing of each ultrasonic sensors and pc imaginative and prescient knowledge being fed into two AI fashions.

All the pieces began with an Arduino Nano ESP32 board linked to a 75KHz ultrasonic sensor and DS18B20 temperature sensor. Aktar then took many readings of his aquarium using the sensor to provide a number of 20×20-point maps, which have been uploaded to the Edge Impulse Studio and used to create a classification mannequin. As soon as educated, the mannequin was deployed to the Nano ESP32 in order that it might inform the person if there have been any toxin-containing bubbles current within the tank.

The opposite half of the monitoring system consists of a DFRobot UNIHIKER single-board pc working a RetinaNet mannequin that detects and classifies chemical exams as both sterile, harmful, or polluted. The outcomes from it and the Nano ESP32 are mixed, offered on a big display screen, and despatched to customers through a Telegram bot.

Extra details about the venture will be present in Atari’s detailed write-up on Hackster.io.