<ul><li>A study on package monitoring in industrial applications using embedded deep learning for detecting package location.</li><li>The proposed approach aims to maximize device lifespan by minimizing wake time through a pipeline involving data processing, training, and evaluation of a deep learning model.</li><li>The method utilizes a one-dimensional Convolutional Neural Network to classify accelerometer data from an IoT device, addressing imbalanced, multiclass time series data.</li><li>Through data augmentation, training, and compression techniques, the model achieves high precision rates for different package states and reduces model size by a factor of four, operating with low power consumption on the IoT device.</li></ul>

Event Classification of Accelerometer Data for Industrial Package Monitoring with Embedded Deep Learning

Discover more