Human-Computer Interaction (HCI) is a sub-area within computer science focused on the study of the communication between people (users) and computers and the evaluation, implementation, and design of user interfaces for computer systems. HCI has accomplished effective incorporation of the human factors and software engineering of computing systems through the methods and concepts of cognitive science. Usability is an aspect of HCI dedicated to guaranteeing that human–computer communication is, amongst other things, efficient, effective, and sustaining for the user. Simultaneously, Human activity recognition (HAR) aim is to identify actions from a sequence of observations on the activities of subjects and the environmental conditions. The vision-based HAR study is the basis of several applications involving health care, HCI, and video surveillance. This article develops a Fire Hawk Optimizer with Deep Learning Enabled Activity Recognition (FHODL-AR) on HCI driven usability. In the presented FHODL-AR technique, the input images are investigated for the identification of different human activities. For feature extraction, a modified SqueezeNet model is introduced by the inclusion of few bypass connections to the SqueezeNet among Fire modules. Besides, the FHO algorithm is utilized as a hyperparameter optimization algorithm, which in turn boosts the classification performance. To detect and categorize different kinds of activities, probabilistic neural network (PNN) classifier is applied. The experimental validation of the FHODL-AR technique is tested using benchmark datasets, and the outcomes reported the improvements of the FHODL-AR technique over other recent approaches.
Usability and Human-Computer Interaction (HCI) are main aspects of the system development processes for enhancing and improving system facilities and for satisfying necessities and needs of users [
Human activity recognition (HAR) serves an important role in human-to-human communication and inter-personal relationships. The reason behind offering information regarding the identity of an individual, their psychological state, and personality, it becomes tough for extraction [
This article develops a Fire Hawk Optimizer with Deep Learning Enabled Activity Recognition (FHODL-AR) on HCI driven usability. In the presented FHODL-AR technique, the input images are investigated for the identification of different human activities. For feature extraction, a modified SqueezeNet model is introduced by the inclusion of few bypass connections to the SqueezeNet among Fire modules. Besides, the FHO algorithm is utilized as a hyperparameter optimization algorithm, which in turn boosts the classification performance. To detect and categorize different kinds of activities, probabilistic neural network (PNN) classifier is applied. The experimental validation of the FHODL-AR technique is tested using benchmark datasets.
Much prevailing research concentrates on feature extraction approaches due to the discriminative characteristics being significant to ensuring the generalizing ability of the HAR mechanism. There were 2 principal ways for extracting features from sensor-related data. One uses hand-crafted features related to the statistical knowledge; another one automatically derives features utilizing neural networks (NN) [
Ronald et al. [
Komang et al. [
In this article, a new FHODL-AR technique has been developed for activity recognition on HCI driven usability. In the presented FHODL-AR technique, the input images are investigated for the identification of different human activities. The overall block diagram is shown in
For feature extraction, a modified SqueezeNet model is introduced by the inclusion of few bypass connections to the SqueezeNet among Fire modules [
To increase the detection performance, an adapted SqueezeNet is developed by including bypass connection to the SqueezeNet among Fire modules. In this framework, bypass connection is additional nearby Fire modules 3, 5, 7, and 9, necessitating the module to learn a residual function among inputs and outputs. To perform a bypass connection around Fire3, we fixed the input to Fire4 equivalent to the output of Fire2 + output of Fire3, whereby the + operator is a component-wise calculation. These variations of the regularization employed to the parameter of the Fire module and, according to ResNet, might enhance the concluding accurateness or trainability.
Here, the FHO algorithm is utilized as a hyperparameter optimization algorithm, which in turn boosts the classification performance [
Now,
Let
In
In
Then, the motion of prey inside the territory of Fire Hawk assumed a major aspect of animal activities for the location updating method.
In
In addition, the prey motion toward the other Fire Hawk territories while there is a chance the prey might be approaching the Fire Hawk in the nearby ambushes or hide in a safe position outside the Fire Hawk territories where they are trapped:
In
Now,
To detect and categorize different kinds of activities, PNN classifier is applied [
In
In
The input unit is distribution unit that supplies a similar input value to the pattern unit and creates a dot product of pattern vector
Then, the non-linear function is a similar method to a Parzen estimator with a Gaussian kernel. The summation units sum the output for pattern unit respective to the class and evaluate the PDF. The output pattern applied the maximum vote to forecast the target class. Because the input layer is applied to the connection weight, PNN doesn’t require altering the connection weight. As a result, the training speed is quicker when compared to the conventional backpropagation neural network (BP-NN).
This section inspects the HAR outcomes of the FHODL-AR model on two datasets. The first KTH dataset (
Label | Class | No. of samples |
---|---|---|
C-1 | Boxing | 100 |
C-2 | Handclapping | 100 |
C-3 | Handwaving | 100 |
C-4 | Jogging | 100 |
C-5 | Running | 100 |
C-6 | Walking | 100 |
Label | Class | No. of samples |
---|---|---|
C-1 | Diving-Side | 100 |
C-2 | Golf-Swing | 100 |
C-3 | Kicking-Front | 100 |
C-4 | Lifting | 100 |
C-5 | Riding Horse | 100 |
C-6 | Run-Side | 100 |
C-7 | StateBoarding-Front | 100 |
C-8 | Swing-Bench | 100 |
C-9 | Swing-SideAngle | 100 |
C-10 | Walk-Front | 100 |
The confusion matrices produced by the FHODL-AR model on KTH dataset are portrayed in
Labels | Accuracy | Precision | Recall | Specificity | F-score |
---|---|---|---|---|---|
C-1 | 99.79 | 98.73 | 100.00 | 99.75 | 99.36 |
C-2 | 98.96 | 95.24 | 98.77 | 99.00 | 96.97 |
C-3 | 98.96 | 100.00 | 93.59 | 100.00 | 96.69 |
C-4 | 98.75 | 94.05 | 98.75 | 98.75 | 96.34 |
C-5 | 99.58 | 100.00 | 97.47 | 100.00 | 98.72 |
C-6 | 99.79 | 100.00 | 98.81 | 100.00 | 99.40 |
C-1 | 98.33 | 95.45 | 95.45 | 98.98 | 95.45 |
C-2 | 100.00 | 100.00 | 100.00 | 100.00 | 100.00 |
C-3 | 99.17 | 95.65 | 100.00 | 98.98 | 97.78 |
C-4 | 100.00 | 100.00 | 100.00 | 100.00 | 100.00 |
C-5 | 99.17 | 100.00 | 95.24 | 100.00 | 97.56 |
C-6 | 100.00 | 100.00 | 100.00 | 100.00 | 100.00 |
C-1 | 96.90 | 87.69 | 91.94 | 97.77 | 89.76 |
C-2 | 97.62 | 95.71 | 90.54 | 99.13 | 93.06 |
C-3 | 97.86 | 95.16 | 90.77 | 99.15 | 92.91 |
C-4 | 99.52 | 98.63 | 98.63 | 99.71 | 98.63 |
C-5 | 98.10 | 90.91 | 98.59 | 97.99 | 94.59 |
C-6 | 97.14 | 93.15 | 90.67 | 98.55 | 91.89 |
C-1 | 99.44 | 100.00 | 97.37 | 100.00 | 98.67 |
C-2 | 98.33 | 92.59 | 96.15 | 98.70 | 94.34 |
C-3 | 99.44 | 97.22 | 100.00 | 99.31 | 98.59 |
C-4 | 99.44 | 96.43 | 100.00 | 99.35 | 98.18 |
C-5 | 98.33 | 96.43 | 93.10 | 99.34 | 94.74 |
C-6 | 99.44 | 100.00 | 96.00 | 100.00 | 97.96 |
The training accuracy (TA) and validation accuracy (VA) achieved by the FHODL-AR algorithm on KTH dataset is displayed in
The training loss (TL) and validation loss (VL) acquired by FHODL-AR process on KTH dataset are shown in
Methods | Accuracy (%) |
---|---|
FHODL-AR | 99.44 |
RNN model | 98.62 |
DBN model | 97.72 |
CNN model | 98.60 |
Bi-LSTM model | 98.19 |
GRU model | 97.05 |
The confusion matrices produced by the FHODL-AR technique on UCF Sports data are indicated in
Labels | Accuracy | Precision | Recall | Specificity | F-score |
---|---|---|---|---|---|
C-1 | 99.38 | 95.89 | 97.22 | 99.59 | 96.55 |
C-2 | 99.50 | 98.73 | 96.30 | 99.86 | 97.50 |
C-3 | 99.38 | 97.78 | 96.70 | 99.72 | 97.24 |
C-4 | 99.38 | 97.44 | 96.20 | 99.72 | 96.82 |
C-5 | 99.50 | 96.51 | 98.81 | 99.58 | 97.65 |
C-6 | 98.50 | 89.87 | 94.67 | 98.90 | 92.21 |
C-7 | 99.00 | 95.29 | 95.29 | 99.44 | 95.29 |
C-8 | 99.75 | 98.68 | 98.68 | 99.86 | 98.68 |
C-9 | 99.12 | 96.30 | 95.12 | 99.58 | 95.71 |
C-10 | 98.75 | 94.52 | 92.00 | 99.45 | 93.24 |
C-1 | 99.00 | 100.00 | 92.86 | 100.00 | 96.30 |
C-2 | 99.00 | 94.74 | 94.74 | 99.45 | 94.74 |
C-3 | 99.00 | 88.89 | 88.89 | 99.48 | 88.89 |
C-4 | 99.00 | 95.24 | 95.24 | 99.44 | 95.24 |
C-5 | 98.50 | 88.24 | 93.75 | 98.91 | 90.91 |
C-6 | 99.50 | 100.00 | 96.00 | 100.00 | 97.96 |
C-7 | 99.50 | 93.75 | 100.00 | 99.46 | 96.77 |
C-8 | 99.50 | 100.00 | 95.83 | 100.00 | 97.87 |
C-9 | 99.00 | 90.00 | 100.00 | 98.90 | 94.74 |
C-10 | 99.00 | 96.00 | 96.00 | 99.43 | 96.00 |
C-1 | 97.86 | 90.48 | 86.36 | 99.05 | 88.37 |
C-2 | 98.29 | 91.04 | 91.04 | 99.05 | 91.04 |
C-3 | 98.14 | 88.10 | 96.10 | 98.39 | 91.93 |
C-4 | 98.86 | 96.88 | 91.18 | 99.68 | 93.94 |
C-5 | 98.86 | 92.75 | 95.52 | 99.21 | 94.12 |
C-6 | 99.00 | 100.00 | 90.54 | 100.00 | 95.04 |
C-7 | 98.71 | 90.67 | 97.14 | 98.89 | 93.79 |
C-8 | 98.86 | 94.20 | 94.20 | 99.37 | 94.20 |
C-9 | 99.43 | 97.01 | 97.01 | 99.68 | 97.01 |
C-10 | 98.29 | 92.00 | 92.00 | 99.04 | 92.00 |
C-1 | 98.33 | 93.94 | 91.18 | 99.25 | 92.54 |
C-2 | 98.00 | 88.57 | 93.94 | 98.50 | 91.18 |
C-3 | 99.33 | 95.65 | 95.65 | 99.64 | 95.65 |
C-4 | 98.33 | 100.00 | 84.38 | 100.00 | 91.53 |
C-5 | 97.67 | 90.62 | 87.88 | 98.88 | 89.23 |
C-6 | 99.33 | 92.86 | 100.00 | 99.27 | 96.30 |
C-7 | 99.33 | 96.67 | 96.67 | 99.63 | 96.67 |
C-8 | 98.00 | 90.32 | 90.32 | 98.88 | 90.32 |
C-9 | 98.67 | 91.43 | 96.97 | 98.88 | 94.12 |
C-10 | 99.67 | 96.15 | 100.00 | 99.64 | 98.04 |
Methods | Accuracy (%) |
---|---|
FHODL-AR | 99.10 |
RNN model | 97.60 |
DBN model | 97.74 |
CNN model | 98.26 |
Bi-LSTM model | 97.01 |
GRU model | 98.21 |
Then, the DBN technique has accomplished somewhat better
In this article, a new FHODL-AR technique has been developed for activity recognition on HCI driven usability. In the presented FHODL-AR technique, the input images are investigated for the identification of different human activities. For feature extraction, a modified SqueezeNet model is introduced by the inclusion of few bypass connections to the SqueezeNet among Fire modules. Besides, the FHO algorithm is utilized as a hyperparameter optimization algorithm, which in turn boosts the classification performance. To detect and categorize different kinds of activities, PNN classifier is applied. The experimental validation of the FHODL-AR technique is tested using benchmark datasets, and the outcomes reported the improvements of the FHODL-AR technique over other recent approaches with maximum accuracy of 99.10%. In the future, the detection performance of the FHODL-AR technique can be boosted by the use of ensemble fusion approaches.