Modern architectures convolutional neural networks in human activity recognition

Recognition of human activity based on convolutional neural network has received the interest of researchers in recent years due to its significant improvement in accuracy. A large number of algorithms based on the deep learning approach have been proposed for activity recognition purpose. However, with the increasing advancements in technologies having limited computational resources, it needs to design an efficient deep learning-based approaches with improved utilization of computational resources. This paper presents a simple and efficient 2-Dimensional convolutional neural network (2-D CNN) architecture with very small size convolutional kernel for human activity recognition. The merit of the proposed CNN architecture over standard deep learning architectures is fewer trainable parameters and lesser memory requirement which enables it to train the proposed CNN architecture on low GPU memory-based devices and also works well with smaller as well as larger size datasets. The proposed approach consists of mainly four stages: namely (1) creation of dataset and data augmentation, (2) designing 2-D convolutional neural network (CNN) architecture, (3) the proposed 2-D CNN architecture trained from scratch up to optimum stage, and (4) evaluation of the trained 2-D CNN architecture. To illustrate the effectiveness of the proposed architecture several extensive experiments are conducted on three publicly available datasets, namely IXMAS, YouTube, and UCF101 dataset. The results of the proposed method and its comparison with other state-of-the-art methods [8-12,14,18-26,29-33] demonstrate the usefulness of the proposed method.

Every year, new Convolutional Neural Network (CNN) architectures appear to deal with different problems in the activity of image and video recognition. These architectures usually work along the ImageNet dataset for looking for the best performance of the CNNs without taking into account the video task where they are used. This can represent a problem if the task is Human Action Recognition (HAR) in video, since the CNN architectures are pre-trained with an image dataset that can practically contain any object, while HAR problem requires consecutive frames of people doing actions. To prove the idea that using CNNs pre-trained on an image dataset does not always achieve the best performance on a video dataset and that, therefore, it is worth comparing the performance of different CNNs under similar circumstances for the HAR problem, this work proposes an analysis between eight different CNN architectures. Each one of the CNN was exclusively trained with RGB images, which were extracted from the frames of the different classes of videos of HMDB51 dataset. To make the classification of an activity in video, we average the predictions taking into account the successes. We also made some ensembles with the best performance CNNs to measure the improvement in accuracy. Our results suggest that Xception is a strong baseline model that could be used by the community to make their comparisons of their proposals more robust.

Hyperparameter optimization poses a significant challenge when developing deep neural networks. Building a convolutional neural network (CNN) for implementation can be an arduous and time-intensive task. This work proposed an approach to optimize the hyperparameters of one dimensional (1D-CNN) to improve the accuracy of human activity recognition (HAR). The framework includes a parametric depiction of 1D-CNNs along with an optimization process for hyperparameters aimed at maximizing the model's performance. This work designed the method called OPTConvNet for hyperparameter optimization of 1D-CNN using Hierarchical Particle Swarm Optimization (H-PSO). The H-PSO algorithm is designed to optimize the architectural, layer and training parameters of 1D-CNN. The H-PSO optimizes the architecture of the 1D-CNN at initial level. Layer and training hyperparameters will be optimized at the next level. The proposed approach employs an exponential-like inertia weight to fine-tune the balance between exploration and exploitation of particles to prevent premature convergence to a local optimum solution in the PSO algorithm. The H-PSO- CNN is evaluated on publicly available sensor- human activity recognition (S-HAR) datasets namely, UCI-HAR, Daphnet Gait, Opportunity and PAMPA2 datasets.

Edge computing aims to integrate computing into everyday settings, enabling the system to be context-aware and private to the user. With the increasing success and popularity of deep learning methods, there is an increased demand to leverage these techniques in mobile and wearable computing scenarios. In this paper, we present an assessment of a deep human activity recognition system's memory and execution time requirements, when implemented on a mid-range smartphone class hardware and the memory implications for embedded hardware. This paper presents the design of a convolutional neural network (CNN) in the context of human activity recognition scenario. Here, layers of CNN automate the feature learning and the influence of various hyper-parameters such as the number of filters and filter size on the performance of CNN. The proposed CNN showed increased robustness with better capability of detecting activities with temporal dependence compared to models using statistical machine learning techniques. The model obtained an accuracy of 96.4% in a five-class static and dynamic activity recognition scenario. We calculated the proposed model memory consumption and execution time requirements needed for using it on a mid-range smartphone. Per-channel quantization of weights and per-layer quantization of activation to 8-bits of precision post-training produces classification accuracy within 2% of floating-point networks for dense, convolutional neural network architecture. Almost all the size and execution time reduction in the optimized model was achieved due to weight quantization. We achieved more than four times reduction in model size when optimized to 8-bit, which ensured a feasible model capable of fast on-device inference.

Mobile fitness applications provide the opportunity to show users real-time feedback on their current fitness activity. For such applications, it is essential to accurately track the user's current fitness activity using available mobile sensors, such as inertial measurement units (IMUs). Convolutional neural networks (CNNs) have been shown to produce strong results in different time series classification tasks, including the recognition of daily living activities. However, fitness activities can present unique challenges to the human activity recognition task (HAR), including greater similarity between individual activities and fewer available data for model training. In this paper, we evaluate the applicability of CNNs to the fitness activity recognition task (FAR) using IMU data and determine the impact of input data size and sensor count on performance. For this purpose, we adapted three existing CNN architectures to the FAR task and designed a fourth CNN variant, which we call the scaling fully convolutional network (Scaling-FCN). We designed a preprocessing pipeline and recorded a running exercise data set with 20 participants, in which we evaluated the respective recognition performances of the four networks, comparing them with three traditional machine learning (ML) methods commonly used in HAR. Although CNN architectures achieve at least 94% test accuracy in all scenarios, two traditional ML architectures surpass them in the default scenario, with support vector machines (SVMs) achieving 99.00 ± 0.34% test accuracy. The removal of all sensors except one foot sensor reduced the performance of traditional ML architectures but improved the performance of CNN architectures on our data set, with our Scaling-FCN reaching the highest accuracy of 99.86 ± 0.11% on the test set. Our results suggest that CNNs are generally well suited for fitness activity recognition, and noticeable performance improvements can be achieved if sensors are dropped selectively, although traditional ML architectures can still compete with or even surpass CNNs when favorable input data are utilized.

Human activity recognition with smartphone sensors using deep learning neural networks

Human activity recognition (HAR) has attracted considerable research attention in the past decade with the development of wearable sensor technology and deep learning algorithms. However, most of the existing HAR methods ignored the spatial relationship of features, which may lead to recognition errors. In this paper, a novel model based on a modified capsule network (MCN) is proposed to accurately recognize various human activities. This novel model is composed of a convolution block and a capsule block, which can achieve end-to-end intelligent recognition. In the meantime, the spatial information among features is preserved through a dynamic routing process. To validate the effectiveness of the model, a human activity dataset is constructed by placing an inertial measurement unit (IMU) on the calf of the volunteers to collect their activity data in daily life, including walking, jogging, upstairs, downstairs, up-ramps, and down-ramps. The recognition accuracy of this novel approach can reach 96.08%, which performs better than the convolutional neural network (CNN) with an accuracy of 91.62%. In addition, it is evaluated on two public datasets named WISDM and UCI-HAR, and the accuracies achieve 98.21% and 95.28%, respectively, which presents higher accuracy than the reported results obtained from benchmark algorithms like CNN. The experimental results show that the proposed model has better activity detection capability and achieves outstanding performance for HAR.

Recently, deep learning, which are able to extract automatically features from data, has achieved state-of-the-art performance across a variety of sensor based human activity recognition (HAR) tasks. However, the existing deep neural networks are usually trained with a global loss, and all hidden layer weights have to be always kept in memory before the forward and backward pass has completed. The backward locking phenomenon prevents the reuse of memory, which is a crucial limitation for wearable activity recognition. In the paper, we proposed a layer-wise convolutional neural networks (CNN) with local loss for the use of HAR task. To our knowledge, this paper is the first that uses local loss based CNN for HAR in ubiquitous and wearable computing arena. We performed experiments on five public HAR datasets including UCI HAR dataset, OPPOTUNITY dataset, UniMib-SHAR dataset, PAMAP dataset, and WISDM dataset. The results show that local loss works better than global loss for tested baseline architectures. At no extra cost, the local loss can approach the state-of-the-arts on a variety of HAR datasets, even though the number of parameters was smaller. We believe that the layer-wise CNN with local loss can be used to update the existing deep HAR methods.

In this paper, we study the performance of variants of well-known Convolutional Neural Network (CNN) architectures on different audio tasks. We show that tuning the Receptive Field (RF) of CNNs is crucial to their generalization. An insufficient RF limits the CNN's ability to fit the training data. In contrast, CNNs with an excessive RF tend to over-fit the training data and fail to generalize to unseen testing data. As state-of-the-art CNN architectures - in computer vision and other domains - tend to go deeper in terms of number of layers, their RF size increases and therefore they degrade in performance in several audio classification and tagging tasks. We study well-known CNN architectures and how their building blocks affect their receptive field. We propose several systematic approaches to control the RF of CNNs and systematically test the resulting architectures on different audio classification and tagging tasks and datasets. The experiments show that regularizing the RF of CNNs using our proposed approaches can drastically improve the generalization of models, out-performing complex architectures and pre-trained models on larger datasets. The proposed CNNs achieve state-of-the-art results in multiple tasks, from acoustic scene classification to emotion and theme detection in music to instrument recognition, as demonstrated by top ranks in several pertinent challenges (DCASE, MediaEval).

Convolutional neural networks (CNN) have shown remarkable performance in enormous computer vision applications over the years, and many works have been done for human activity recognition (HAR) using CNN. However, most of deep learning-based architectures require large training data and plenty of computational resources. Therefore, we proposed a simple and efficient deep learning model for human activity recognition, which works on complex visual data and require less computational resources. In the proposed work, we designed a novel CNN architecture by stacking the repeated components called micro-networks to incorporate multiscale processing in the network layers. We have used a feature fusion strategy to pass the previous layer’s abstract complementary information to the next adjacent layer, believing that each layer encapsulates specific feature maps. Therefore, the hidden complementary information can potentially enhance the feature discrimination capacity of the network and help in learning the network. The proposed architecture is fully trained from scratch with stochastic gradient descent (SGD) optimizer at 0.05 initial learning rate and a softmax classifier is used for activity recognition. The merit of the proposed method over standard deep learning models is its computational efficiency in terms of learnable parameters and computational resources. The proposed model gives good performance on small as well as large size datasets. For authentication of the proposed method several extensive experiments are conducted on publically available datasets, namely UCF-101, HMDB-51, YouTube, and IXMAS datasets. The results have shown the outperformance of the proposed method over the existing state-of-the-art methods.

In this paper, the Convolutional Neural Network (CNN) architecture, which is one of the deep learning architectures, is used to classify the basic circuit components drawn by hand. During the training and testing stages of the model, a new dataset containing images of 863 circuit components manually drawn by different people is created. The data set contains images of four different classes of circuit components such as resistor, inductor, capacitor and voltage source. All images have been fixed to the same size and converted to grayscale to increase recognition performance and reduce process complexity. In the study, training for four classes is performed with CNN architecture. Based on the CNN architecture, four new CNN models are employed with different the number of layers. The training and validation results of these models are compared separately, the model with the highest training and validation performance is observed with four layer CNN model (CNN-4). This model obtained 84.41% accuracy rate at classification task.

Human activity recognition (HAR) can play a vital role in biomedical and health informatics by enabling the monitoring of human daily activities and health behaviors. Accurate HAR can provide valuable insights into patients' physical activity levels, and this helps to manage chronic conditions and promote healthy lifestyles. In this paper, we propose a deep learning model, DKInception, designed for HAR tasks. DKInception integrates deep convolutional residual networks with an attention mechanism and leverages multi-scale convolution kernels to efficiently extract temporal features for activity identification. This model is built on the Inception ResNet architecture and extends its capabilities with effective, fast convergence and robust scaling properties. To evaluate the performance of DKInception, we conduct extensive experiments on four benchmark HAR datasets: UCI-HAR, Opportunity, Daphnet, and PAMAP2. Our comparative analysis with several existing models shows that DKInception outperforms these models using various evaluation metrics. These results demonstrate the model recorded high accuracy of 95.70%, 87.48%, 94.00% and 89.72%, for UCI-HAR, Opportunity, Daphnet, and PAMAP2, respectively.

In recent years, channel state information (CSI) in WiFi 802.11n has been increasingly used to collect data pertaining to human activity. Such raw data are then used to enhance human activity recognition. Activities such as lying down, falling, walking, running, sitting down, and standing up can now be detected with the use of information collected through CSI. Human activity recognition has a multitude of applications, such as home monitoring of patients. Four deep learning models are presented in this paper, namely: a convolution neural network (CNN) with a Gated Recurrent Unit (GRU); a CNN with a GRU and attention; a CNN with a GRU and a second CNN, and a CNN with Long Short-Term Memory (LSTM) and a second CNN. Those models were trained to perform Human Activity Recognition (HAR) using CSI amplitude data collected by a CSI tool. Experiments conducted to test the efficacy of these models showed superior results compared with other recent approaches. This enhanced performance of our models may be attributable the ability of our models to make full use of available data and to extract all data features, including high dimensionality and time sequence. The highest average recognition accuracy reached by the proposed models was achieved by the CNN-GRU, and the CNN-GRU with attention models, standing at 99.31% and 99.16%, respectively. In addition, the performance of the models was evaluated for unseen CSI data by training our models using a random split-of-dataset method (70% training and 30% testing). Our models achieved impressive results with accuracies reaching 100% for nearly all activities. For the lying down activity, accuracy obtained from the CNN-GRU model stood at 99.46%; slightly higher than the 99.05% achieved by the CNN-GRU with attention model. This confirmed the robustness of our models against environmental changes.

Covid-19 first occurred in Wuhan, China in December 2019. Subsequently, the virus spread throughout the world and as of June 2020 the total number of confirmed cases are above 4.7 million with over 315,000 deaths. Machine learning algorithms built on radiography images can be used as a decision support mechanism to aid radiologists to speed up the diagnostic process. The aim of this work is to conduct a critical analysis to investigate the applicability of convolutional neural networks (CNNs) for the purpose of COVID-19 detection in chest X-ray images and highlight the issues of using CNN directly on the whole image. To accomplish this task, we use 12-off-the-shelf CNN architectures in transfer learning mode on 3 publicly available chest X-ray databases together with proposing a shallow CNN architecture in which we train it from scratch. Chest X-ray images are fed into CNN models without any preprocessing to replicate researches used chest X-rays in this manner. Then a qualitative investigation performed to inspect the decisions made by CNNs using a technique known as class activation maps (CAM). Using CAMs, one can map the activations contributed to the decision of CNNs back to the original image to visualize the most discriminating region(s) on the input image. We conclude that CNN decisions should not be taken into consideration, despite their high classification accuracy, until clinicians can visually inspect and approve the region(s) of the input image used by CNNs that lead to its prediction.

Although font recognition is a fundamental issue in the field of document analysis and recognition, it was usually ignored in the past. With the development of optical character recognition (OCR), font recognition becomes more and more important. This paper proposed a well-designed convolutional neural network (CNN) architecture for traditional Mongolian font recognition by means of a single word. To be specific, the whole word image is regarded as input of CNN. Hence, the word images should be normalized into the same size before being inputted into CNN. By comparison, an appropriate aspect ratio for the traditional Mongolian word images has been determined. Experimental results demonstrate that the proposed CNN architecture outperforms three classic CNN architectures, including LeNet-5, AlexNet and GoogLeNet. Therefore, the proposed CNN is much more suitable for the task of the traditional Mongolian font recognition in the way of a single word.