Motion-compensated online object tracking for activity detection and crowd behavior analysis

Patel, Ashish Singh; Vyas, Rushi; Vyas, O. P.; Ojha, Muneendra; Tiwari, Vivek

doi:10.1007/s00371-022-02469-3

Cited by 16 publications

(4 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Patel [18] presented a multi-object tracking algorithm to generate reliable object tracks for examining crowd behavior in public places by reducing the occlusion of short-term objects, identity switches, and detection errors. The detection of the bounding box and velocity estimation of a linear object utilizing the Kalman filter were used to track the object's frame by frame and the missing detections and temporary object occlusion were handled by maintaining the predicted tracks in existence.…”

Section: Literature Surveymentioning

confidence: 99%

Abnormal Crowd Behaviour Detection in Surveillance Videos Using Spatiotemporal Inter-Fused Autoencoder

2023

IJIES

View full text Add to dashboard Cite

In crowded environments, the importance of automatic video surveillance cannot be overstated. It plays a vital role in detecting unusual incidents and averting accidents, particularly in areas teeming with pedestrians. Surveillance frameworks demonstrate their inherent strength when deployed in real-world scenarios. In this paper, an efficient algorithm that can identify abnormalities in videos has been proposed as a solution to the problems of high computing power and variation between negative and positive samples. To address the issue of fewer negative samples, the algorithm employs the spatiotemporal inter-fused autoencoder in a method of unsupervised learning to locate and extract the samples of negative from the dataset. A spatial-temporal convolutional neural network (CNN) is built using this methodology, and it has a basic structure and requires minimal computing power. To develop the model of detection, the spatiotemporal CNN is trained using the method of supervised training with negative and positive data. The UMN and UCSD datasets are used as benchmark datasets for this method. The outcomes of the experiment demonstrate that the proposed method is more accurate than the existing algorithms at both frame and pixel levels and it can locate anomalous behaviors in real-time. The proposed method achieves better accuracy of 99.76%, 99.92%, 99.15%, 98.45%, and 94.67% in UCSD ped1, UCSD ped2, and UMN scene 1, 2, 3 datasets compared to the hybrid CNN and RF classifiers, MCMS-BCN attention+densenet121/Efficientnetv2, gradient motion descriptor (PGD) and enhanced entropy classifier, and attention mechanism. Additionally, the proposed method achieves better AUC of 99.96%, 99.83%, 99.97%, 90.15%, and 99.72% in the datasets of UCSD ped1, UCSD ped2, and UMN scenes 1, 2, 3 compared to the low-rank and compact coefficient dictionary learning (LRCCDL), and hybrid CNN and RF classifiers.

show abstract

Section: Literature Surveymentioning

confidence: 99%

Abnormal Crowd Behaviour Detection in Surveillance Videos Using Spatiotemporal Inter-Fused Autoencoder

2023

IJIES

View full text Add to dashboard Cite

show abstract

“…MOT combines multiple disciplines, such as pattern recognition, machine learning, computer vision, image processing, and computer applications, forming a technology for multi-object localisation and motion trajectory prediction. MOT technology has broad application prospects and tremendous potential economic value in fields, such as intelligent surveillance, behaviour analysis, human-computer interaction, sports analysis, and intelligent driving systems [5][6][7][8][9].…”

Section: Introductionmentioning

confidence: 99%

A deep learning framework for multi‐object tracking in team sports videos

Cao,

Wang,

Liu

et al. 2024

IET Computer Vision

View full text Add to dashboard Cite

In response to the challenges of Multi‐Object Tracking (MOT) in sports scenes, such as severe occlusions, similar appearances, drastic pose changes, and complex motion patterns, a deep‐learning framework CTGMOT (CNN‐Transformer‐GNN‐based MOT) specifically for multiple athlete tracking in sports videos that performs joint modelling of detection, appearance and motion features is proposed. Firstly, a detection network that combines Convolutional Neural Networks (CNN) and Transformers is constructed to extract both local and global features from images. The fusion of appearance and motion features is achieved through a design of parallel dual‐branch decoders. Secondly, graph models are built using Graph Neural Networks (GNN) to accurately capture the spatio‐temporal correlations between object and trajectory features from inter‐frame and intra‐frame associations. Experimental results on the public sports tracking dataset SportsMOT show that the proposed framework outperforms other state‐of‐the‐art methods for MOT in complex sport scenes. In addition, the proposed framework shows excellent generality on benchmark datasets MOT17 and MOT20.

show abstract

“…It is possible to spot patterns and trends that can shed light on the scene's underlying dynamics by focusing attention on how objects behave and communicate. Analyzing historical and current motion patterns allows for predicting future movements in a given scene, a capability essential for applications like traffic management, crowd control, and public safety [8]. Motion patterns in crowded scenes can be categorized into three primary types: structured, unstructured, and semi-structured.…”

Section: Introductionmentioning

confidence: 99%

Motion Pattern-Based Scene Classification Using Adaptive Synthetic Oversampling and Fully Connected Deep Neural Network

Mohammed,

Al-Dhamari,

Saeed

et al. 2023

IEEE Access

View full text Add to dashboard Cite

Analyzing crowded environments has become an increasingly researched topic in computer vision community, largely due to its myriad practical applications, including enhanced video surveillance systems and the estimation of crowd density in specific settings. This paper presents a comprehensive method for progressing the study of crowd dynamics and behavioral analysis, specifically focusing on the classification of movement patterns. We introduce a specialized neural network-based classifier explicitly designed for the accurate categorization of various crowd scenes. This classifier fills a unique niche in the existing literature by offering robust and adaptive classification capabilities. To optimize the performance of our model, we conduct an in-depth analysis of loss functions commonly employed in multi-class classification tasks. Our study encompasses four widely used loss functions: Focal Loss, Huber Loss, Cross-Entropy Loss, and Multi-Margin Loss. Based on empirical findings, we introduce a Joint Loss function that combines the strengths of Cross-Entropy and Multi-Margin Loss, outperforming existing methods across key performance metrics such as accuracy, precision, recall, and F1-score. Furthermore, we address the critical challenge of class imbalance in motion patterns within crowd scenes. To this end, we perform a comprehensive comparative study of two leading oversampling techniques: the synthetic minority oversampling technique (SMOTE) and adaptive synthetic sampling (ADASYN). Our results indicate that ADASYN is superior at enhancing classification performance. This approach not only mitigates the issue of class imbalance but also provides robust empirical validation for our proposed method. Finally, we subject our model to a rigorous evaluation using the Collective Motion Database, facilitating a comprehensive comparison with existing state-of-the-art techniques. This evaluation confirms the effectiveness of our model and aligns it with established paradigms in the field.

show abstract

Motion-compensated online object tracking for activity detection and crowd behavior analysis

Cited by 16 publications

References 61 publications

Abnormal Crowd Behaviour Detection in Surveillance Videos Using Spatiotemporal Inter-Fused Autoencoder

Abnormal Crowd Behaviour Detection in Surveillance Videos Using Spatiotemporal Inter-Fused Autoencoder

A deep learning framework for multi‐object tracking in team sports videos

Motion Pattern-Based Scene Classification Using Adaptive Synthetic Oversampling and Fully Connected Deep Neural Network

Contact Info

Product

Resources

About