Título: Previsão do nível de um tanque em tempo real utilizando IA embarcada em um ESP32 -CAM -S3 através da plataforma Edge Impulse e TensorFlow lite
Autores: Gustavo Wanderley, Wellington Silva de Oliveira, Diego Medeiros Ponte, Daniel Lopes Martins, Heitor Medeiros Florencio & Adrião Duarte Dória Neto
Resumo: This paper presents the implementation of an embedded computer vision model on the ESP32-CAM microcontroller to measure a tank’s level. The system uses the Edge Impulse platform for model training and TensorFlow Lite for its execution. The model, based on convolutional neural networks, was quantized to ensure efficiency on resource-constrained devices. The AI system’s performance is rigorously evaluated through a direct comparison with a high-precision industrial level sensor, validating its accuracy and identifying its limitations in a practical scenario.
Palavras-chave: IA; GPU; Quantização; ESP32; Visão Computacional; Edge Impulse; TensorFlow Lite.
Páginas: 7
Código DOI: 10.21528/CBIC2025-1191495
Artigo em PDF: CBIC_2025_paper1191495.pdf
Arquivo BibTeX:
CBIC_2025_1191495.bib