This is the companion repository for our paper iSPLInception: Redefining the State-of-the-Art for Human Activity Recognition which will be published in IEEE Access - 2021.
Advances in deep learning model design have pushed the boundaries of the areas in which it can be applied. The fields with an immense availability of complex big data have been big beneficiaries of these advances. One such field is human activity recognition (HAR). HAR is a popular area of research in a connected world because internet-of-things (IoT) devices and smartphones are becoming more prevalent. The goal of recent research work has been to improve predictive accuracy for devices with limited computational resources. In this paper, we propose iSPLInception, a deep learning model motivated by the Inception-ResNet architecture from Google, that redefines the state-of-the-art for HAR. We evaluate the proposed model’s performance on four public HAR datasets from the University of California, Irvine (UCI) machine learning repository. The proposed model’s classification performance is compared to that of existing deep learning architectures that have been proposed in the recent past to solve the HAR problem. The proposed model outperforms these existing approaches based on several metrics that include accuracy, cross-entropy loss, and the F1 score on all four datasets. This paper establishes a verifiable state-of-the-art benchmark for the UCI HAR using smartphones dataset, Opportunity activity recognition dataset, Daphnet freezing of gait dataset, and PAMAP2 physical activity monitoring dataset that were retrieved from the UCI repository.
This project and repository contain a Python project whose objective is to