ESP-Fi HAR: A low-power WiFi CSI dataset for Ad-Hoc IoT human activity recognition

We introduces a new 3-D video dataset to assess the performance of Human Activity Recognition system in indoor and outdoor environment. This dataset also help to check the performance of activity recognition algorithms against the effect of varying illumination, background and viewpoint. The available dataset for activity recognition are simple and most of them contain RGB information only as well as fewer complex and far away from real world general scenario. This dataset consist of RGB and depth information for all the activities performed under different illumination condition and viewpoint. This dataset includes ten activities performed by eleven subjects in four illumination condition and three viewpoints. Around one hundred and ten videos of one hour duration are recorded and annotated. We believe that the additional depth information provided will be useful for researcher in analyzing the performance of activity recognition algorithms for real time implementation. Large no of dataset are publicly available but most of them are less complex and consist of RGB data only. The proposed dataset is more complex and comprises of depth information along with RGB.

A survey of video datasets for human action and activity recognition

Currently, open data and data sets are emerging in human activity recognition (HAR) due to their importance in different application areas such as improving people's lives, enabling informed care decisions, real-world problem solutions, and strategies for choosing the best HAR approaches. There are challenges associated with curating and sharing open data and data sets due to the absence of metadata and complete descriptions of the shared data. By properly curating data sets it will be easier to recognise, obtain and reuse to help make progress in HAR research. In this paper, we propose a conceptual framework for understanding the open data set lifecycle as consisting of four phases of construction, sharing, finding, and using. Similarly, open issues and challenges are explored related to HAR data sets from the published literature. On this basis, an approach is presented to automatically extract metadata through web scraping of the HAR data sets and then perform a natural language processing (NLP) pipeline to detect the metadata of data sets. As a result of metadata retrieval, we show how comparisons can be performed under different scenarios which can help evaluate data set quality and identify areas for improvement in data set curation. This research work will assist the HAR research community in better understanding the open data set lifecycle and how data set quality can be improved.

Human action recognition is a critical capability for our survival, allowing us to interact easily with the environment and others in everyday life. Although the neural basis of action recognition has been widely studied using a few action categories from simple contexts as stimuli, how the human brain recognizes diverse human actions in real-world environments still needs to be explored. Here, we present the Human Action Dataset (HAD), a large-scale functional magnetic resonance imaging (fMRI) dataset for human action recognition. HAD contains fMRI responses to 21,600 video clips from 30 participants. The video clips encompass 180 human action categories and offer a comprehensive coverage of complex activities in daily life. We demonstrate that the data are reliable within and across participants and, notably, capture rich representation information of the observed human actions. This extensive dataset, with its vast number of action categories and exemplars, has the potential to deepen our understanding of human action recognition in natural environments.

Aeriform in-action: A novel dataset for human action recognition in aerial videos

Human activity recognition can help in elderly care by monitoring the physical activities of a subject and identifying a degradation in physical abilities. Vision-based approaches require setting up cameras in the environment, while most body-worn sensor approaches can be a burden on the elderly due to the need of wearing additional devices. Another solution consists in using smart glasses, a much less intrusive device that also leverages the fact that the elderly often already wear glasses. In this article, we propose UCA-EHAR, a novel dataset for human activity recognition using smart glasses. UCA-EHAR addresses the lack of usable data from smart glasses for human activity recognition purpose. The data are collected from a gyroscope, an accelerometer and a barometer embedded onto smart glasses with 20 subjects performing 8 different activities (STANDING, SITTING, WALKING, LYING, WALKING_DOWNSTAIRS, WALKING_UPSTAIRS, RUNNING, and DRINKING). Results of the classification task are provided using a residual neural network. Additionally, the neural network is quantized and deployed on the smart glasses using the open-source MicroAI framework in order to provide a live human activity recognition application based on our dataset. Power consumption is also analysed when performing live inference on the smart glasses’ microcontroller.

Domain shift is a common problem in many real-world applications using machine learning models. Most of the existing solutions are based on supervised and deep-learning models. This paper proposes a novel clustering algorithm capable of producing an adapted and/or integrated clustering model for the considered domains. Source and target domains are represented by clustering models such that each cluster of a domain models a specific scenario of the studied phenomenon by defining a range of allowable values for each attribute in a given data vector. The proposed domain integration algorithm works in two steps: (i) cross-labeling and (ii) integration. Initially, each clustering model is crossly applied to label the cluster representatives of the other model. These labels are used to determine the correlations between the two models to identify the common clusters for both domains, which must be integrated within the second step. Different features of the proposed algorithm are studied and evaluated on a publicly available human activity recognition (HAR) data set and real-world data from a smart logistics use case provided by an industrial partner. The experiment's goal on the HAR data set is to showcase the algorithm's potential in automatic data labeling. While the conducted experiments on the smart logistics use case evaluate and compare the performance of the integrated and two adapted models in different domains.

Recent advancement in smart phones and computing technologies has played a vital role in people’s life. Develop a model to detect the human basic dynamic activities such as Amble, Climb stairs, coming down the stairs into the floor and human basic static activities like Sitting, Standing or Laying using the person’s smart phone and computers are the major work of this paper. Conventional Machine learning models like Logistic Regression, SVC, Decision tree, etc. results are compared with a recurrent deep neural network model named as Long Short Term Memory (LSTM). LSTM is proposed to detect the human behavior based on Human Activity Recognition (HAR) dataset. The data is monitored and recorded with the aid of sensors like accelerometer and Gyroscope in the user smart phone. HAR dataset is collected from 30 persons, performing different activities with a smart phone to their waists. The testing of the model is evaluated with respect to accuracy and efficiency. The designed activity recognition system can be manipulated in other activities like predicting abnormal human actions, disease by human actions, etc. The overall accuracy has improved to 95.40%.

This work introduces the harAGEdataset: a novel multimodal smartwatch-based dataset for Human Activity Recognition (HAR) with more than 17 hours of data collected from 19 participants using a Garmin Vivoactive 3 device. The dataset contains samples from resting, lying, sitting, standing, washing hands, walking, running, stairs climbing, strength workout, flexibility workout, and cycling activities. The resting activity, excluded from the set of activities to recognise, was explicitly conducted while avoiding stressors and external stimuli, so the data collected can be used to compute the personal, baseline heart rate at rest. We also present the HAR-based models trained using the accelerometer data to recognise different sets of activities. Specifically, we focus on different strategies to combine, fuse, and enrich the accelerometer measurements, so they can be used end-to-end. Model performances are assessed following a Leave-One-Subject-Out Cross-Validation (LOSO-CV) approach, and we use the Unweighted Average Recall (UAR) as the evaluation metric to compare the ground truth and the inferred information. The best UAR score of 98.1 % is obtained when recognising the static and the dynamic activities, excluding the samples corresponding to the washing hands, strength workout, and flexibility workout activities. When recognising the specific activities included in these two sets, the model with the best performance scores a UAR of 70.1 %. Finally, when recognising all the activities considered in the harAGEdataset, the highest UAR achieved is 64.3 %.

Human action recognition is an active research topic in both computer vision and machine learning communities, which has broad applications including surveillance, biometrics and human computer interaction. In the past decades, although some famous action datasets have been released, there still exist limitations, including the limited action categories and samples, camera views and variety of scenarios. Moreover, most of them are designed for a subset of the learning problems, such as single-view learning problem, cross-view learning problem and multi-task learning problem. In this paper, we introduce a multi-view, multi-modality benchmark dataset for human action recognition (abbreviated to MMA). MMA consists of 7080 action samples from 25 action categories, including 15 single-subject actions and 10 double-subject interactive actions in three views of two different scenarios. Further, we systematically benchmark the state-of-the-art approaches on MMA with respective to all three learning problems by different temporal-spatial feature representations. Experimental results demonstrate that MMA is challenging on all three learning problems due to significant intra-class variations, occlusion issues, views and scene variations, and multiple similar action categories. Meanwhile, we provide the baseline for the evaluation of existing state-of-the-art algorithms.

Activity recognition from human action data is quite a challenging task in the biomedical data science community. The main challenge in dealing with human activity recognition (HAR) datasets is their high cardinality. Therefore, reducing cardinality is a cardinal area of research in the HAR field. In this research, reducing the data dimensionality by utilizing future selection methods has been used. This research work has extracted features using wavelet packet transform (WPT) and the cardinality of the feature set has been reduced by using the Genetic Algorithm (GA) technique. The selected features also have been ranked according to their importance based on their SHAP values. In the venture, an interesting inspection has been found. That is in HAR datasets, signal values lay into lower frequency regions mostly. The highest accuracy and f1-score which have been got are 94.74%, 94.73%, and 89.98%, 89.67% for the feature extracted and feature selected dataset respectively.

Objectives: This study is an endeavor to provide quick, on-the-go classification of a human activity dataset with an aim to improve on the classification time of a machine learning algorithm for Human Activity Recognition (HAR) datasets. Methods/Statistical analysis: It proposes the use of a customized sampler called the Normal On-The-Go (Normal OTG) sampler to reduce the classification time. Concocted using a combination of stratified, random and normal sampling, the Normal OTG sampler was tested on HAR datasets and was found to significantly reduce the training time of the most commonly used machine learning algorithms. Three datasets, ShoaibSA, ShoaibPA and USC-HAD were used to conduct the experiments. Findings: It was found that using as little as 5% samples from the training dataset sampled by the Normal OTG sampler, sufficiently reliable accuracy was obtained from most of the 9 classifiers that were used. The results indicated that almost 96% of time was saved in the training process in the case of USC-HAD, and 62% and 83% time was saved in the case of ShoaibPA and ShoaibSA respectively. It was also found that the results were consistent among the three datasets. Application/Improvements: The study helps training of data in human activity recognition a faster process and thereof, making algorithm selection a less tedious procedure Keywords: Classification Time, Human Activity Recognition, Robust, Sampling Technique

Human activity recognition is attracting increasing research attention. Many activity recognition datasets have been created to support the development and evaluation of new algorithms. Given the lack of datasets collected in real environments (In The Wild) to support human activity recognition in public spaces, we introduce a large-scale video dataset for activity recognition In The Wild: POLIMI-ITW-S. The fully labeled dataset consists of 22,161 RGB video clips (about 46 h) including 37 activity classes performed by 50 K+ subjects in real shopping malls. We evaluated the state-of-the-art models on this dataset and get relatively low accuracy. We release the dataset including the annotations composed by person tracking bounding boxes, 2-D skeleton, and activity labels for research use at: https://airlab.deib.polimi.it/polimi-itw-s-a-shopping-mall-dataset-in-the-wild.

Human activity recognition has a significant impact on people’s daily lives. The need to infer human activities is prominent in many human-centric applications, such as healthcare and individual assistance. In this article, we introduce a noninvasive human activity recognition system that utilizes footstep-induced vibration and sound in an outdoor environment with the aim of achieving improved performance over a single source of information. We employ 1-D convolutional neural networks (1-D CNNs) for automated feature extraction, fusion, and activity recognition on a nine-class classification problem. The proposed framework reports an average F1 score of 92%, which corresponds to a 5.74% improvement over the best-performing state-of-the-art. Confusion matrix-based analysis demonstrates that audio-seismic fusion not only reduces misclassifications, but also reduces the impact of background noise on model performance. In addition, we demonstrate that a model trained on a balanced dataset has a higher F1 score than one trained on an imbalanced dataset. Activity-wise performance is reported to show the efficacy of the proposed fusion-based framework. We also contribute an audio-seismic dataset for human activity recognition in an outdoor environment. The dataset is collected in a variety of challenging environments, such as varying grass length, soil moisture content, and the passing of unwanted vehicles.

Human activity recognition (HAR) has made a significant impact in a variety of applications, not least of which includes wearable Internet of Things (IoT). Considerable research is currently underway in the field of machine learning, deep learning and neural networks, an objective of which is to improve the efficiency of relevant algorithms, as well as to increase the truthfulness of the recognition model. For purposes of analysis and consistency, much of the published research uses either self-collected sensor-based databases or publicly available databases for potential research objectives. There exist diverse open source sensor-based time-series datasets that are applied to the testing of algorithm efficiency, which is then used for human activity detection. Essential to algorithm performance testing are datasets that consist of balanced data and labels. Such datasets vary in size, labeling, accuracy, number of activities performed by each user, type of sensor data included, and various other elements. In this paper, we evaluate two popular datasets by comparing selected factors; we also demonstrate our experimental results. We implement minimal data processing by extracting five basic features - mean, standard deviation, absolute mean error, mean squared error, and histogram plots – and perform experiments on both datasets to study their outcomes. We compare the UCI HAR and WISDM (Wireless Sensor Data Mining) datasets, based on their accuracy, precision, recall, and F1-score on nine different machine learning algorithms. Both datasets contain tri-axial sensor data. Experimental results show that the UCI HAR dataset is the more promising of the two. Additionally, we show that with basic data processing, simple activity recognition is possible with current state-of-the-art algorithms. Specifically, as we increase the type of sensor data, the accuracy of simple human activity detection is improved. We conclude with a discussion of results, as well as study limitations, challenges, and future research.