Design and Implementation of an Automatic Plant Watering Device with 2D Visual Gesture Recognition

This report presents the development of an advanced automatic plant watering device, which integrates both autonomous operation and 2D visual gesture recognition capabilities. The device features a robust embedded system architecture designed for seamless interaction with sensors and external systems, including MQTT for real-time communication. The hardware design incorporates a NodeMCU ESP8266, soil moisture sensors, a BME280 sensor, an OLED display, and a servo mechanism for water delivery. The system architecture emphasizes non-blocking task execution and extensibility, utilizing a custom scheduler and MQTT client to manage tasks such as sensor data collection and plant watering routines. External integrations with Node Red and MQTT.fx provide additional functionality for monitoring and control. The report also details the computer vision component, which uses OpenCV for gesture recognition from webcam feeds, enabling user interaction through visual commands. Performance evaluations of the gesture recognition system, including segmentation and hand tracking, are discussed. This comprehensive approach ensures effective plant care while offering intuitive user interaction via visual gestures.