Skip to main content
Have a personal or library account? Click to login
Revolutionizing Big Data Assessment for Human Activity Recognition with Metapath Context and Bi-directional Cascade Networks Cover

Revolutionizing Big Data Assessment for Human Activity Recognition with Metapath Context and Bi-directional Cascade Networks

Open Access
|Sep 2026
Read online

Figures & tables

Table 1.

Comparison of the existing work with HAR

RefMethodsProsConsPerformance measures used
[10]TD-NetThe method was effective in both dimensions, delivering strong performance on extensive datasets.It failed to address the computational complexity associated with implementation, limiting its practical application in scenarios requiring immediate processing.Accuracy-79.3%
F1-score-64.15%
Training time-0.5ms
Inference time-0.05ms
Params-1.88M
GFLOPs-0.028
[11]CNN-RNNThe DT model enhances the descriptive human action features by integrating the DWT technique. This approach improves the performance of action recognition by leveraging both engineered features and learned ones, thereby harnessing the advantages of both methodologies.Absence of temporal domain diversity in terms of time scales.Accuracy-95.68
[12]CNN-BiLSTMThe approach was intended to seize both short-range and extended dependencies within sequential data, ensuring a comprehensive understanding of temporal relationships across various time scales.Inability to incorporate contextual information from multiple branches into the model architecture hindered its capability to effectively seize nuanced outlines in the data.Accuracy-96.37
F1-score-96.31
Training time-53.68
Params-631,014
[13]HAR transformerGreater temporal spacing between features in the time series provided enhanced contextual comprehension across extended sequences.Lack of testing the modified transformer on a prolonged database, preferably integrating diverse sensor data, limited a comprehensive evaluation.Accuracy-99.2
Precision-99.2
Recall-99
[14]CNN-LSTMThe recognition accuracy was improved by leveraging user similarity and weighting training data, allowing for a more refined learning process that accounted for individual user characteristics and preferences.It failed to focus on the advances observed in wearable and phone-based tracking systems.Accuracy-98.76
Params-634,052
Recall-99.93
[15]ViCo-MoCo-DLThis method effectively utilizes video encoding and motion estimation techniques, combined with DL, for enhanced HAR.Absence of evaluation across diverse datasets limits the generalizability of the findings.Accuracy-97%
[16]CNN-Bi-GRUThis technique leverages DL through CNN and Bi-GRUs for improved HAR performance.Lack of comprehensive feature selection techniques, theoretically leading to suboptimal performance and increased computational complexity.Accuracy-93.38
Average time-7.15s
[17]Deep BiLSTMThe utilization of the model improved through transfer-learning-based attribute extraction for dynamic HAR.Inability to evaluate across diverse datasets, potentially limiting the generalizability of the model’s performance.Accuracy-99.2%
[18]TransTMBy utilizing a time-streaming multiscale transformer for device-free HAR.Lack of extensive validation across diverse scenarios, potentially limiting the generalizability of its performance outside controlled environments.Accuracy-99.1%
F1-score-99.1%
FLOPs-27.98M
Params-1.82M
[19]RNN-LSTMIts capability to perform DL based HAR directly on edge devices.Inability to constrain resources in edge environments, potentially limiting the model’s performance or scalability on such platforms.Accuracy-99.31%
Recall-99.09%
F1-score-99.06%
Time-0.298ms
Figure 1.

Workflow of the proposed work

Figure 2.

MobileNetV2 with GCN model

Table 2.

Pseudo code of EFTTO

Initialize FTTA
CLC
  // Function parameters objectives set
   Set Maxgen = 500 // Iterations numbers
   Set npop = 50 // Population size
   Set Pstudy = 0.2 // Probability of study
   Set Pcomm = 0.2 // Probability of communication
   Set Perror = 0.001 // Probability of error
  // Initialization
   For I = 1 to Maxgen Do
   // Initialization loop starts here
 For each i from 1 to npop Do
 The fitness value of each football player is calculated, which is expressed in Eqn. (11):
 End for
 Find the finest player & the worst player
 // Collective Training Phase begins
 Collective training:
(13)
Hi,jknew={Hi,jkold+rand∗(Hbest,jk−Hi,jkold)Hi,jkold+rand1∗(Hbest,jk−Hi,jkold)−rand2∗(Hworst,jk−Hi,jkold)Hi,jkold+rand∗(Hbest,jk−Hworst,jk)Hi,jkold∗(1+K(t))
(14)
Hi,jk,team1new={ Hbest,jk,team1if rand≤pstudyHrandom,jk,team1if rand≤pstudyHrandom,j∗(1+randn)k,team1if rand≤Pcomm
(15)
Hi,jk,team1new=Hrandom1,random2k,team1if rand≤perror
For every i : 1 ≤ i ≤ nPop do
End for
Update the players
Find the best player
Additional individualized training:
(16)
Hbestknew={ Hbestkold∗(1+(1k)∗gauss+1k∗cauchy)if H(Hbestknew)<H(Hbestkold)Hbestkold if H(Hbestknew)>H(Hbestkold)
End for
T = T + 1
Exit the while loop
Exit the for loop
Figure 3.

Confusion matrix of the corresponding datasets

Figure 4.

(a, b and c). Class-wise accuracies of the corresponding datasets

Figure 5.

(a), (b), (c) and (d). Activity recognition outcomes using the proposed method

Table 3.

Accuracy analysis comparison of the JHMDB dataset

TechniquesAccuracy (%)
TD-net [10]79.3
MF + MVF [13]87.76
ViCo-MoCo [15]78.5
Deep BiLSTM [17]76.3
MCCBCN (Proposed)99.34
Table 4.

Accuracy analysis comparison of the HMDB51 dataset

TechniquesAccuracy (%)
CNN-RNN [11]70.8
MF + MVF [13]71.28
ViCo-MoCo [15]71.4
CNN Bi-GRU [16]71.89
MCCBCN (Proposed)99.45
Figure 6.

(a) accuracy, (b) precision, (c) recall, and (d) F1-score on recent HAR techniques

Figure 7.

Time complexity analysis

Figure 8.

K-fold validation analysis

Figure 9.

Fitness curve

Table 5.

Ablation study of the introduced method with their datasets

Model ConfigurationJHMDB Accuracy (%)UCF 11 Accuracy (%)UCF YouTube Action Accuracy (%)HMDB51 Accuracy (%)
Full MCCBCN architecture99.3499. 3999.4299.45
Without Metapath Context85.986.785.985.3
Without Bi-directional Cascade89.386.287.788.3
Without EFTTOA Integration89.188.188.587.2
Without LSMOST88.585.586.187.8
Without MobileNetV2-Lite & GCN Integration88.988.186.787.7
DOI: https://doi.org/10.14313/jamris-2026-047 | Journal eISSN: 2080-2145 | Journal ISSN: 1897-8649
Language: English
Page range: 155 - 169
Submitted on: Nov 5, 2024
Accepted on: May 20, 2025
Published on: Sep 30, 2026
Published by: Łukasiewicz Research Network – Industrial Research Institute for Automation and Measurements PIAP
In partnership with: Paradigm Publishing Services
Publication frequency: 4 issues per year

© 2026 Praveen S. Banasode, Sunita Padmannavar, published by Łukasiewicz Research Network – Industrial Research Institute for Automation and Measurements PIAP
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.