AI-Driven Salient Soccer Events Recognition Framework for Next-Generation IoT-Enabled Environments

Muhammad, Khan; Ullah, Hayat; Obaidat, Mohammad S.; Ullah, Amin; Munir, Arslan; Sajjad, Muhammad; Albuquerque, Victor Hugo C. de

doi:10.1109/jiot.2021.3110341

Cited by 11 publications

(4 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They first extracted deep discriminative CNN features from video frames using VGG19, which are then fed into LSTM to analyze hidden sequential patterns and recognize human activities. Muhammad et al [6] presented a spatio-termporal approach for recognizing salient events in soccer videos, where they used a pretrained ResNet50 architecture for deep feature extraction and a multilayer LSTM for event recognition from the hidden sequential patterns. These hybrid CNN+LSTM approaches exhibited significant performance for vision-based human action-and activity-recognition tasks; these methods are computationally complex due to intensive computation caused by CNN feature extraction and human action modeling by LSTM.…”

Section: Temporal Modeling-based Methodsmentioning

confidence: 99%

“…For instance, running involves rapid movement of hands and legs; similarly, throwing object involves the backward and forward force of arms and hands. Human activity recognition has numerous potential applications, such as in smart surveillance systems [3], video summarization [4], content-based video retrieval [5], sports and healthcare [6], and human-computer interactions [7]. In video, each frame contributes spatial information in sequential order which forms a sequential pattern containing human activity that cannot be recognized in a single video frame.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Human Activity Recognition Using Cascaded Dual Attention CNN and Bi-Directional GRU Framework

Ullah

Munir

2023

J. Imaging

Self Cite

View full text Add to dashboard Cite

Vision-based human activity recognition (HAR) has emerged as one of the essential research areas in video analytics. Over the last decade, numerous advanced deep learning algorithms have been introduced to recognize complex human actions from video streams. These deep learning algorithms have shown impressive performance for the video analytics task. However, these newly introduced methods either exclusively focus on model performance or the effectiveness of these models in terms of computational efficiency, resulting in a biased trade-off between robustness and computational efficiency in their proposed methods to deal with challenging HAR problem. To enhance both the accuracy and computational efficiency, this paper presents a computationally efficient yet generic spatial–temporal cascaded framework that exploits the deep discriminative spatial and temporal features for HAR. For efficient representation of human actions, we propose an efficient dual attentional convolutional neural network (DA-CNN) architecture that leverages a unified channel–spatial attention mechanism to extract human-centric salient features in video frames. The dual channel–spatial attention layers together with the convolutional layers learn to be more selective in the spatial receptive fields having objects within the feature maps. The extracted discriminative salient features are then forwarded to a stacked bi-directional gated recurrent unit (Bi-GRU) for long-term temporal modeling and recognition of human actions using both forward and backward pass gradient learning. Extensive experiments are conducted on three publicly available human action datasets, where the obtained results verify the effectiveness of our proposed framework (DA-CNN+Bi-GRU) over the state-of-the-art methods in terms of model accuracy and inference runtime across each dataset. Experimental results show that the DA-CNN+Bi-GRU framework attains an improvement in execution time up to 167× in terms of frames per second as compared to most of the contemporary action-recognition methods.

show abstract

Section: Temporal Modeling-based Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Human Activity Recognition Using Cascaded Dual Attention CNN and Bi-Directional GRU Framework

Ullah

Munir

2023

J. Imaging

Self Cite

View full text Add to dashboard Cite

show abstract

“…Many universities and schools do not have the resources to buy these costlier machines and technologies and maintain them continuously. These institutes may need extra funding to incorporate them into their teaching practices in the classrooms [12]. Left apart from the concerns related to security, privacy, and the job demand market is also there.…”

Section: Evaluating the Key Issues Faced By Learners Of The Chinese L...mentioning

confidence: 99%

AI-Assisted Chinese Language Teaching System in Distributed Sensor Networks

Shan

2023

SCPE

View full text Add to dashboard Cite

The Chinese learning system gained enormous importance to support the population of China with recent technologies. The AI and ML-assisted classes are useful for learners. However, it can cut down the employment of teachers. Various planning and act was passed to reform the Chinese education system with artificial intelligence and machine language. On the other hand, this education process is interesting, customized, and unsupervised based. It was observed that the human touch study process is more satisfying and clears doubts easily. The key object of data gathering sensor-based network is accumulating all the data from the nodes and sending those sensed data to the base station. These nodes are basically working while focusing on one thought: all the nodes present in the sensor-based system are well aware of all the other nodes currently working in the same network and transmitting data in the base station. Although AI comes with a few challenges like lack of resources, training, and various security issues, if these are overcome with proper strategies, AI can be the equipment to reign in the 21${}^{st}$ century. Additionally, LLMs and NLP modeling are described in the results that helped to understand the use of different models. Moreover, the Benefits of LLMs and NLP for Chinese language training are discussed in the study.

show abstract

“…They have first extracted deep discriminative CNN features from video frames using VGG19, which are then fed into LSTM for analyzing hidden sequential patterns and recognition of human activities. Muhammad et al [44] have presented a spatio-termporal approach for recognizing salient events in soccer videos, where they have used a pretrained ResNet50 architecture for deep features extraction and a multilayer LSTM for events recognition from the hidden sequential patterns. Although, these hybrid CNN+LSTM have shown significant performance for vision-based human action and activity recognition task, these methods are computationally complex due to intensive computation cause by CNN features extraction and human action modeling by LSTM.…”

Section: Related Work On Human Activity Recognitionmentioning

confidence: 99%

Human Activity Recognition Using Cascaded Dual Attention CNN and Bi-Directional GRU Framework

Ullah¹,

Munir²

2022

Preprint

Self Cite

View full text Add to dashboard Cite

<p>Vision-based human activity recognition has emerged as one of the essential research areas in video analytics domain. Over the last decade, numerous advanced deep learning algorithms have been introduced to recognize complex human actions from video streams. These deep learning algorithms have shown impressive performance for the human activity recognition task. However, these newly introduced methods either exclusively focus on model performance or the effectiveness of these models in terms of computational efficiency and robustness, resulting in a biased tradeoff in their proposals to deal with challenging human activity recognition problem. To overcome the limitations of contemporary deep learning models for human activity recognition, this paper presents a computationally efficient yet generic spatial-temporal cascaded framework that exploits the deep discriminative spatial and temporal features for human activity recognition. For efficient representation of human actions, we have proposed an efficient dual attentional convolutional neural network (CNN) architecture that leverages a unified channel-spatial attention mechanism to extract human-centric salient features in video frames. The dual channel-spatial attention layers together with the convolutional layers learn to be more attentive in the spatial receptive fields having objects over the number of feature maps. The extracted discriminative salient features are then forwarded to stacked bi-directional gated recurrent unit (Bi-GRU) for long-term temporal modeling and recognition of human actions using both forward and backward pass gradient learning. Extensive experiments are conducted on three publicly available human action datasets, where the obtained results verify the effectiveness of our proposed framework over the state-of-the-art methods in terms of model accuracy and inference runtime across each dataset. Experimental results show that the proposed framework attains an improvement in execution time up to 167× in terms of frames per second as compared to most of the contemporary action recognition methods. </p>

show abstract

AI-Driven Salient Soccer Events Recognition Framework for Next-Generation IoT-Enabled Environments

Cited by 11 publications

References 36 publications

Human Activity Recognition Using Cascaded Dual Attention CNN and Bi-Directional GRU Framework

Human Activity Recognition Using Cascaded Dual Attention CNN and Bi-Directional GRU Framework

AI-Assisted Chinese Language Teaching System in Distributed Sensor Networks

Human Activity Recognition Using Cascaded Dual Attention CNN and Bi-Directional GRU Framework

Contact Info

Product

Resources

About