Object Tracking Research Articles

Keypoint prediction enhanced Siamese networks with attention for accurate visual object tracking

Keypoint prediction enhanced Siamese networks with attention for accurate visual object tracking

Using Fuzzy Clustering Technology to Implementing Multiple Unresolvable Group Object Tracking in a Clutter Environment

Using Fuzzy Clustering Technology to Implementing Multiple Unresolvable Group Object Tracking in a Clutter Environment

Correlation-Embedded Transformer Tracking: A Single-Branch Framework.

Developing robust and discriminative appearance models has been a long-standing research challenge in visual object tracking. In the prevalent Siamese-based paradigm, the features extracted by the Siamese-like networks are often insufficient to model the tracked targets and distractor objects, thereby hindering them from being robust and discriminative simultaneously. While most Siamese trackers focus on designing robust correlation operations, we propose a novel single-branch tracking framework inspired by the transformer. Unlike the Siamese-like feature extraction, our tracker deeply embeds cross-image feature correlation in multiple layers of the feature network. By extensively matching the features of the two images through multiple layers, it can suppress non-target features, resulting in target-aware feature extraction. The output features can be directly used to predict target locations without additional correlation steps. Thus, we reformulate the two-branch Siamese tracking as a conceptually simple, fully transformer-based Single-Branch Tracking pipeline, dubbed SBT. After conducting an in-depth analysis of the SBT baseline, we summarize many effective design principles and propose an improved tracker dubbed SuperSBT. SuperSBT adopts a hierarchical architecture with a local modeling layer to enhance shallow-level features. A unified relation modeling is proposed to remove complex handcrafted layer pattern designs. SuperSBT is further improved by masked image modeling pre-training, integrating temporal modeling, and equipping with dedicated prediction heads. Thus, SuperSBT outperforms the SBT baseline by 4.7%,3.0%, and 4.5% AUC scores in LaSOT, TrackingNet, and GOT-10K. Notably, SuperSBT greatly raises the speed of SBT from 37 FPS to 81 FPS. Extensive experiments show that our method achieves superior results on eight VOT benchmarks.

A Pig Behavior-tracking Method Based on a Multi-Channel High-Efficiency Attention Mechanism

Because the behavior of pigs reflects their health status, continuous and precise monitoring of behavior is important for effective health management and welfare protection. To mitigate potential tracking failures during analysis of video footage, we introduced a novel multi-target pig tracking method that consisted of detection and tracking components. The detection model was enhanced with an efficient attention mechanism and a Cross Stage Partial Darknet backbone network, which significantly improved detection accuracy. The tracking component used the Bytetrack algorithm to accurately track the movement trajectories of individual pigs. Together, these components were combined into the Dual-YOLOX-Tiny-ByteTrack (DYTB) model, which demonstrated superior performance in automatic monitoring of pig behaviors compared with previously published approaches. We established multi-object pig tracking datasets with 180,321 images to evaluate this method. The DYTB method achieved a pig detection accuracy of 98.3% and tracking accuracies of 95.3% and 97.1%. Compared with the YOLOX-Tiny-ByteTrack base model, DYTB showed a 3.4% improvement in multiple object tracking accuracy, making it a robust method for non-contact, intelligent monitoring of pig health that can contribute to advances in precision livestock farming.

A Robust Visual Tracking Method based on Correlation Filter

Abstract Although the fast discriminative scale space tracking(fDSST) method shows superior performance for short time visual tracking, it is prone to tracking failure when the target is occluded or moving fast in case of long time tracking. To address this issue, we proposed a novel enhanced visual tracking method based on fDSST for robust tracking. Specifically, based on correlation filter response map we design a visual tracking status discrimination method by integrating Peak to Sidelobe Ratio(PSR) and the number of response peaks. Then, we design an adaptive model update method coupled with extended search area strategy to reduce the probability of target loss. Extensive experiments are performed on challenging benchmark sequences from Online Object Tracking Benchmark(OTB) with significant target occlusion and fast motion. Ours results show that the proposed approach improves the DP by 11.9% and AUC by 8.4% compared to the baseline fDSST, and operates at real-time.

MVP-HOT: A moderate visual prompt for hyperspectral object tracking

The growing attention to hyperspectral object tracking (HOT) can be attributed to the extended spectral information available in hyperspectral images (HSIs), especially in complex scenarios. This potential makes it a promising alternative to traditional RGB-based tracking methods. However, the scarcity of large hyperspectral datasets poses a challenge for training robust hyperspectral trackers using deep learning methods. Prompt learning, a new paradigm emerging in large language models, involves adapting or fine-tuning a pre-trained model for a specific downstream task by providing task-specific inputs. Inspired by the recent success of prompt learning in language and visual tasks, we propose a novel and efficient prompt learning method for HOT tasks, termed Moderate Visual Prompt for HOT (MVP-HOT). Specifically, MVP-HOT freezes the parameters of the pre-trained model and employs HSIs as visual prompts to leverage the knowledge of the underlying RGB model. Additionally, we develop a moderate and effective strategy to incrementally adapt the HSI prompt information. Our proposed method uses only a few (1.7M) learnable parameters and demonstrates its effectiveness through extensive experiments, MVP-HOT can achieve state-of-the-art performance on three hyperspectral datasets.

Scene-aware classifier and re-detector for thermal infrared tracking

Compared with common visible light scenes, the target of infrared scenes lacks information such as the color, texture. Infrared images have low contrast, which not only lead to interference between targets, but also interference between the target and the background. In addition, most infrared tracking algorithms lack a redetection mechanism after lost target, resulting in poor tracking effect after occlusion or blurring. To solve these problems, we propose a scene-aware classifier to dynamically adjust low, middle, and high level features, improving the ability to utilize features in different infrared scenes. Besides, we designed an infrared target re-detector based on multi-domain convolutional network to learn from the tracked target samples and background samples, improving the ability to identify the differences between the target and the background. The experimental results on VOT-TIR2015, VOT-TIR2017 and LSOTB-TIR show that the proposed algorithm achieves the most advanced results in the three infrared object tracking benchmark.

A Lightweight Robust RGB-T Object Tracker Based on Jitter Factor and Associated Kalman Filter

Visual object tracking has made significant contributions in many practical applications, but it remains a great challenge when the camera moves/shakes or the target is occluded. Various solutions leveraging deep-learning (DL) techniques have been introduced to address these challenging factors. However, these DL-based methods can hardly be implemented on an edge computing platform due to its limited computational resources. In this study, we propose a lightweight and robust cross-modal fusion RGB-T object tracker for edge computing platforms based on Jitter Factor and associated Kalman filter. In the proposed tracker, visible and infrared features of the target are extracted and fused using a cross-modal fusion strategy based on the modal reliability. Meanwhile, the newly proposed Jitter Factor, derived from image morphology, is used to judge the motion of camera. Once the camera motion is detected, target position would be corrected via global image registration and associated Kalman filter. Experimental results on RGBT234 and GTOT datasets indicates that the proposed lightweight tracking method outperforms other non-DL-based tracking methods. For the problem of camera motion, it exhibits a competitive performance among other DL-based trackers, but with faster speed (25 FPS using only a CPU).

Local-Global Self-Attention for Transformer-Based Object Tracking

Local-Global Self-Attention for Transformer-Based Object Tracking

ATN plasma biomarkers and digital cognitive performance in adults with type 1 diabetes: data from the GluCog study

AbstractBackgroundHighly specific ATN plasma biomarker assays for neurodegenerative diseases have been developed, but their associations with cognition vary in different populations. Kidney disease, common in diabetes, may decrease the predictive precision of those biomarkers. The aim of this study was to characterize for the first time the relationships between plasma ATN biomarkers and cognitive function in adults with T1D.MethodAdults with T1D mean age 49 years (range 19‐84), 53% female, mean HbA1c 7.6% who participated in the Glycemic Variability and Fluctuations in Cognitive Status in Adults with T1D study and had plasma β‐amyloid42/40, p‐tau181, neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP) levels measured using Quanterix, and p‐tau217 using AlzPath assays, were included (N=114). Self‐administered online cognitive tests (TestMyBrain.org) were used: Digit Symbol Matching (processing speed), Gradual Onset Continuous Performance (cognitive control, attention), Multiple Object Tracking (visual working memory); Vocabulary (verbal reasoning), Matrix (non‐verbal reasoning), Simple Reaction Time (psychomotor speed), Letter‐Number Switching (cognitive flexibility), Visual Paired Associates (visual memory), Flicker (working memory), and Paced Serial Addition Test (PSAT, working memory, attention). Bivariate correlation analyses between cognitive function and biomarkers were performed. Significant correlations after false discovery rate were re‐analyzed using linear multiple regression with biomarker assigned as outcome and cognitive function as predictors in separate models, adjusting for age, sex and education.ResultAll biomarkers were correlated with cognitive function in bivariate analysis. Multivariate analyses revealed that higher concentrations of ptau181 and GFAP were associated with longer reaction time for correct responses, PSAT (β=0.002, p=0.011 and β=0.057, p=0.012, respectively), and higher β‐amyloid42/40 concentration was associated with better vocabulary performance (β=0.028, p=0.009). Associations remained significant after including kidney‐related disease or other diabetes features as adjustments.ConclusionPTau181, GFAP and β‐amyloid42/40 were associated with cognitive performance in adults with T1D. Working memory speed was related to ptau181 and GFAP, suggesting it may be an early indicator of neurodegeneration in this population. The positive association between β‐amyloid42/40 and Vocabulary, a measure of crystalized cognitive ability, suggests protective mechanisms related to cognitive reserve. Investigation on ATN biomarkers and longitudinal cognitive decline is crucial for disentangling their role in T1D.

The restricted intermittent control for high-speed train movement via the full state dependent event-triggered method.

In this paper, the full state dependent event-triggered aperiodic intermittent control (FE-AIC) strategy based on input constraints is introduced to minimize energy consumption and enhance speed tracking accuracy in the high-speed train (HST) operation. Firstly, a dynamic model based on multi-mass-point (MMP) for HST has been established, transforming the cruise control problem into an error asymptotic convergence problem. Secondly, restricted FE-AIC (RFE-AIC) controller is designed separately in the presence and absence of external disturbances to realize tracking objects. The proposed control scheme is not only based on control input constraints, but also intermittent control with full state event dependence. The RFE-AIC scheme and the conditions for determining parameters are given, which ensures the stability of the ideal tracking speed and coupler deviation at the equilibrium point. Eventually, the availability of the proposed method in cruise control is confirmed through numerical simulations. It is proved that the RFE-AIC has better performance compared with the self-triggered and guaranteed optimal cruise control methods.

A Review of CNN-Based Object Recognition and Tracking Systems for Visually Impaired People

Visually impaired persons (VIPs) face significant challenges in navigating their environments, impacting their independence and quality of life. This paper reviews recent advancements in assistive technologies aimed at enhancing mobility and safety for VIPs. It explores various systems, including wearable devices, object detection technologies, and navigation aids, highlighting their functionalities, limitations, and potential future develop- ment. The integration of machine learning and sensor technologies has led to innovative solutions that provide realtime feedback and improved spatial awareness for users. VIPsneed assistance in performing daily tasks such as object detection, obstacle recognition,and navigation, particularly in indoor and outdoor environments. The paper emphasizes the importance of protecting and ensuring the safety of VIPs. Devices typically use sensors like infrared, ultrasonic, and imagery to gather environmental data, which is then processed by machine learning techniques. Users receive feedback via auditory or vibratory means. The paper provides a comprehensive comparative analysis of assistive devices, discussing their key attributes, challenges, and limitations, along with a score-based evaluation to help select appropriate devices for specific needs

Development of Artificial Intelligent-Based Methodology to Prepare Input for Estimating Vehicle Emissions

Machine learning has significantly advanced traffic surveillance and management, with YOLO (You Only Look Once) being a prominent Convolutional Neural Network (CNN) algorithm for vehicle detection. This study utilizes YOLO version 7 (YOLOv7) combined with the Kalman-based SORT (Simple Online and Real-time Tracking) algorithm as one of the models used in our experiments for real-time vehicle identification. We developed the “ISTraffic” dataset. We have also included an overview of existing datasets in the domain of vehicle detection, highlighting their shortcomings: existing vehicle detection datasets often have incomplete annotations and limited diversity, but our “ISTraffic” dataset addresses these issues with detailed and extensive annotations for higher accuracy and robustness. The ISTraffic dataset is meticulously annotated, ensuring high-quality labels for every visible object, including those that are truncated, obscured, or extremely small. With 36,841 annotated examples and an average of 32.7 annotations per image, it offers extensive coverage and dense annotations, making it highly valuable for various object detection and tracking applications. The detailed annotations enhance detection capabilities, enabling the development of more accurate and reliable models for complex environments. This comprehensive dataset is versatile, suitable for applications ranging from autonomous driving to surveillance, and has significantly improved object detection performance, resulting in higher accuracy and robustness in challenging scenarios. Using this dataset, our study achieved significant results with the YOLOv7 model. The model demonstrated high accuracy in detecting various vehicle types, even under challenging conditions. The results highlight the effectiveness of the dataset in training robust vehicle detection models and underscore its potential for future research and development in this field. Our comparative analysis evaluated YOLOv7 against its variants, YOLOv7x and YOLOv7-tiny, using both the “ISTraffic” dataset and the COCO (Common Objects in Context) benchmark. YOLOv7x outperformed others with a mAP@0.5 of 0.87, precision of 0.89, and recall of 0.84, showing a 35% performance improvement over COCO. Performance varied under different conditions, with daytime yielding higher accuracy compared to night-time and rainy weather, where vehicle headlights affected object contours. Despite effective vehicle detection and counting, tracking high-speed vehicles remains a challenge. Additionally, the algorithm’s deep learning estimates of emissions (CO, NO, NO2, NOx, PM2.5, and PM10) were 7.7% to 10.1% lower than ground-truth.

Vehicle Target Detection Using the Improved YOLOv5s Algorithm

This paper explores the application of the YOLOv5s algorithm integrated with the DeepSORT tracking detection algorithm in vehicle target detection, leveraging its advantages in data processing, loss function, network structure, and training strategy. Regarding detection frame regression, adopting Focal-EIOU can improve vehicle detection accuracy by precisely measuring overlap and better handle complex scenarios, enhancing the overall performance. The CoordConv convolution layer with more spatial position information is employed to enhance the original network structure’s convolution layer and improve vehicle positioning accuracy. The principle and effectiveness of the Shuffle Attention mechanism are analyzed and added to the YOLOv5s network structure to enhance training, improve detection accuracy and running speed. And the DeepSORT tracking detection algorithm is designed to achieve high-speed operation and high-accuracy matching in target tracking, enabling efficient and reliable tracking of objects. Simultaneously, the network structure is optimized to enhance algorithmic speed and performance. To meet the requirements of vehicle detection in practical transportation systems, real-world vehicle images are collected as a dataset for model training to achieve accurate vehicle detection. The results show that the accuracy rate P of the improved YOLOv5s algorithm is increased by 0.484%, and mAP_0.5:0.95 reaches 92.221%, with an increase of 1.747%.

Fast moving table tennis ball tracking algorithm based on graph neural network

The key object tracking in sports video scenarios poses a pivotal challenge in the analysis of sports techniques and tactics. In table tennis, due to the small size and rapid motion of the ball, identifying and tracking the table tennis ball through video is a particularly arduous task, where the majority of existing detection and tracking algorithms struggle to meet the practical application requirements in real-world scenarios. To address this issue, this paper proposes a combined technical approach integrating detection and discrimination, tailored to the unique motion characteristics of table tennis. For the detector, we utilize and refine a common video differential detector. As for the discriminator, we introduce GMP (a Graph Max-message Pass Neural Network), which is designed specifically for tracking table tennis balls or similar objects. Furthermore, we enhance an existing dataset for table tennis tracking problems by enriching its scenarios. The results demonstrate that our proposed technical solution performs impressively on both the dataset and the intended real-world environments, showcasing the good scalability of our algorithms and models as well as their potential for application in other scenarios.

Motion-Aware Self-Supervised RGBT Tracking with Multi-Modality Hierarchical Transformers

Supervised RGBT (SRGBT) tracking tasks need both expensive and time-consuming annotations. Therefore, the implementation of Self-Supervised RGBT (SSRGBT) tracking methods has become increasingly important. Straightforward SSRGBT tracking methods use pseudo-labels for tracking, but inaccurate pseudo-labels can lead to object drift, which severely affects tracking performance. This article proposes a self-supervised RGBT object tracking method (S2OTFormer) to bridge the gap between tracking methods supervised under pseudo-labels and ground truth labels. Firstly, to provide more robust appearance features for motion cues, we introduce a multi-modality hierarchical transformer (MHT) module for feature fusion. This module allocates weights to both modalities and strengthens the expressive capability of the MHT module through multiple nonlinear layers to fully utilize the complementary information of the two modalities. Secondly, in order to solve the problems of motion blur caused by camera motion and inaccurate appearance information caused by pseudo-labels, we introduce a motion-aware mechanism (MAM). The MAM extracts the average motion vectors from the previous multi-frame search frame features and constructs the consistency loss with the motion vectors of the current search frame features. The motion vectors of inter-frame objects are obtained by reusing the inter-frame attention map to predict coordinate positions. Finally, to further reduce the effect of inaccurate pseudo-labels, we propose an Attention-Based Multi-Scale Enhancement Module. By introducing cross-attention to achieve more precise and accurate object tracking, this module overcomes the receptive field limitations of traditional CNN tracking heads. We demonstrate the effectiveness of S2OTFormer on four large-scale public datasets through extensive comparisons as well as numerous ablation experiments. The source code is available at https://github.com/LiShenglana/S2OTFormer .

Correction to: ReSOTS: RFID/IoT-Enabled Secure Object Tracking Key Exchange for Trustworthy Smart Logistics

Correction to: ReSOTS: RFID/IoT-Enabled Secure Object Tracking Key Exchange for Trustworthy Smart Logistics

Radar based Object Detector using Arduino Uno

Radar technology has been the backbone of several applications, ranging from aviation to meteorology, offering extremely valuable capabilities for object detection and tracking. Conventional radar systems, though highly powerful, are very costly and complex, which makes them not easily accessible in many fields. To address these issues, this research attempts to design and implement a radar system using Arduino micro controllers and ultrasonic sensors. This innovative approach relies on the flexibility and cost-effectiveness of the Arduino platform and the accuracy of ultrasonic sensors for distance measurement to build a radar-like system. The work addresses the development of the hardware, acquisition of data, processing of the signal, and real-time visualization, thus offering an accessible and costeffective alternative for applications of radar. The research will integrate Arduino technology with ultrasonic sensors into a radar system; it focuses on simplicity, affordability, and practicality, achieving The area it covers is between traditional radar and do-it-yourself systems, and then application windows open up as robotics, smart cities, industrial automation, and eventually into the education sector as well. The results outline above are opening new visions to designs in radar systems while pointing radar innovations to newer avenues of work

Comparative analysis of visual and cognitive performance in esports athletes and soccer players

ABSTRACT Clinical relevance Assessing visual and cognitive performance in athletes is important for optimising training strategies and preventing visual strain in both esports and traditional sports. Background This study compared visual acuity, refractive error, binocular functions, and cognitive perception measures between esports and soccer players to optimise visual performance strategies. Methods Twenty-eight male esports athletes and 38 male soccer players were assessed. Visual acuity was measured using an ETDRS chart, and refractive error was assessed with retinoscopy. Phoria was evaluated with the Modified Thorington Technique at distance and near. Fusional vergence ranges and vergence facility were measured using prism bars and a prism flipper. Monocular accommodative facility at near was tested with ± 2.00D flipper lenses. Stereopsis was assessed using the Randot Stereo Test. Perception span, multiple object tracking, and visual reaction time, were evaluated using the Senaptec Sensory Station. Results Soccer players had significantly better visual acuity than esports athletes. For the right eye, the soccer group had an average decimal visual acuity of 1.1 ± 0.2 compared to 0.9 ± 0.1 in the esports group (p < 0.001). Refractive error analysis showed that esports athletes had more myopia (M =-0.67 ± 1.70D) and higher astigmatism (J0 = 0.47 ± 0.53D) than soccer players (M = 0.6 ± 1.06D, J0 = 0.09 ± 0.43D; p < 0.001). Soccer players demonstrated better negative fusional vergence at distance (p = 0.049) and superior near-accommodative facility (12.4 ± 4.2 cpm vs. 9.4 ± 5.2 cpm; p = 0.025). Soccer players also outperformed esports athletes in multiple object tracking (1811.3 ± 391.3 vs. 1523.4 ± 528.7; p = 0.013). No significant differences were observed in stereopsis or other binocular vision parameters. Conclusion Soccer players showed superior visual acuity and performance in tasks such as negative fusional vergence and near-accommodative facility compared to esports players. Esports athletes had more myopia and astigmatism, indicating a need for targeted visual interventions.

Electrophysiology Reveals That Intuitive Physics Guides Visual Tracking and Working Memory.

Starting in early infancy, our perception and predictions are rooted in strong expectations about the behavior of everyday objects. These intuitive physics expectations have been demonstrated in numerous behavioral experiments, showing that even pre-verbal infants are surprised when something impossible happens (e.g., when objects magically appear or disappear). However, it remains unclear whether and how physical expectations shape different aspects of moment-by-moment online visual scene processing, unrelated to explicit physical reasoning. In two EEG experiments, people watched short videos like those used in behavioral studies with adults and infants, and more recently in AI benchmarks. Objects moved on a stage, and were briefly hidden behind an occluder, with the scene either unfolding as expected, or violating object permanence (adding or removing an object). We measured the contralateral delay activity, an electrophysiological marker of online processing, to examine participants' working memory (WM) representations, as well as their ability to continuously track the objects in the scene. We found that both types of object permanence violations disrupted tracking, even though violations involved perceptually non-salient events (magical vanishing) or new objects that weren't previously tracked (magical creation). Physical violations caused WM to reset, i.e., to discard the original scene representation before it could recover and represent the updated number of items. Providing a physical explanation for the violations (a hole behind the occluder) restored object tracking, and we found evidence that WM continued to represent items that disappeared 'down the hole'. Our results show how intuitive physical expectations shape online representations, and form the basis of dynamic object tracking.