Optical Waste Identifier Using Google’s Vision AI and MQTT Protocol





Vasconcelos-Magaña, Adrián
Montes-Peña, Luis D.
Martínez-Martínez, Ricardo

Título de la revista

ISSN de la revista

Título del volumen





This project was developed with the intention of providing a tool that helps users to manage waste generated at home. The development was based on the use of two microcontrollers: one in charge of managing the activation of actuators and indicators, as well as the physical interaction with the user. The second was used to carry out the image processing by creating a connection to Google's Vision AI. Both boards were connected wirelessly via the MQTT protocol. The outcome regarding the use of Vision AI helps the reader to understand the ease of use of this tool, as well as its versatility, leading to an easy incorporation in different projects related to image processing. On the other hand, the results referring to the MQTT protocol demonstrate why it should be one of the first considerations when the objective is to create a stable and scalable wireless connection between two or more clients through the same server. The results obtained adduce the basis for future research, which should provide scalability to the project and take it to be a commercial product that will help any user who is seeking to perform greener practices at home.

Palabras clave

Vision AI, Adafruit, MQTT, Google Cloud, Raspberry Pi


Vasconcelos-Magaña, A.; Montes-Peña, L. D.; Martínez-Martínez, R. (2021). Optical Waste Identifier Using Google’s Vision AI and MQTT Protocol. Trabajo de obtención de grado, Especialidad en Sistemas Embebidos. Tlaquepaque, Jalisco: ITESO.