Sequential Modeling for the Recognition of Activities in Logistics
Zafi Sherhan Syed, Muhammad Zaigham Abbas Shah Syed, Muhammad Shehram Shah Syed, Aunsa Shah · Sukkur IBA Journal of Emerging Technologies · 2021
Activity recognition is an important task in cyber physical system research and has been the focus of researchers worldwide. This paper presents a method for activity recognition in logistic operations using data from accelerometer and gyroscope sensors. A Long Short Term Memory (LSTM) recurrent neural network, bidirectional LSTM and a Convolutional LSTM (ConvLSTM) are used to classify between six activities being performed in the logistics operations being carried out. Comparing the performance of the LSTMs to the Conv-LSTM network, the designed Bi-LSTM RNN outperforms the other networks considered