A parallel spatiotemporal saliency and discriminative online learning method for visual target tracking in aerial videos

Aghamohammadi, Amirhossein; Ang, Mei Choo; Sundararajan, Elankovan; Ng, Kok Weng; Mogharrebi, Marzieh; Banihashem, Seyed Yashar

doi:10.1371/journal.pone.0192246

Cited by 20 publications

(7 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…ii) Although the proposed spatio-temporal processing scheme leads to average F-score improvement for all the involved automatic detection algorithms, the minor loss of TPs and the remaining FPs prevented the achievement of an average F-score higher than 0.9, so this should receive further attention. iii) All seven aerial video datasets contain a limited number of image frames, so larger databases should be established so that adequate training data is available to implement object detection models using some machine learning and deep convolutional neural network (CNN) related schemes [34]- [44]; iv) GPU-based parallel computing has great potential for reducing computation time.…”

Section: Conclusion and Further Workmentioning

confidence: 99%

“…As future work, we plan to consider a parallel design for spatio-temporal processing via a high-performance computer [34] for better utilization of memory and, hence, decreased computation time. We also suggest improving the adapted MMA algorithm by adding a frame differencing method as a pre-processing step, then exploiting better multi-scale feature selection to improve average F-score with reference to some of the latest models [35]- [44] on object detection and their post-processing schemes [45].…”

Section: Conclusion and Further Workmentioning

confidence: 99%

“…We also suggest improving the adapted MMA algorithm by adding a frame differencing method as a pre-processing step, then exploiting better multi-scale feature selection to improve average F-score with reference to some of the latest models [35]- [44] on object detection and their post-processing schemes [45]. Furthermore, we are investigating some of the recent deep learning schemes [46]- [57] for detecting and tracking vehicles [27], [34], [39], [58]- [61] in accordance with the complexity analysis [45], [62]- [69] from the deep CNN-based multi-object detection and segmentation schemes [48]- [51], [53]- [56], [59]- [61], [64]- [66], [70]- [72] applied to wide-area aerial surveillance.…”

Section: Conclusion and Further Workmentioning

confidence: 99%

See 2 more Smart Citations

Spatio-Temporal Processing for Automatic Vehicle Detection in Wide-Area Aerial Video

et al. 2020

View full text Add to dashboard Cite

Vehicle detection in aerial videos often requires post-processing to eliminate false detections. This paper presents a spatio-temporal processing scheme to improve automatic vehicle detection performance by replacing the thresholding step of existing detection algorithms with multi-neighborhood hysteresis thresholding for foreground pixel classification. The proposed scheme also performs spatial post-processing, which includes morphological opening and closing to shape and prune the detected objects, and temporal post-processing to further reduce false detections. We evaluate the performance of the proposed spatial processing on two local aerial video datasets and one parking vehicle dataset, and the performance of the proposed spatio-temporal processing scheme on five local aerial video datasets and one public dataset. Experimental evaluation shows that the proposed schemes improve vehicle detection performance for each of the nine algorithms when evaluated on seven datasets. Overall, the use of the proposed spatio-temporal processing scheme improves average F-score to above 0.8, achieves a 90% reduction in false positives and reduces percentage of wrong classifications to below 50% for some of the nine algorithms. INDEX TERMS Vehicle detection, hysteresis thresholding, spatio-temporal processing, wide-area aerial imagery.

show abstract

Section: Conclusion and Further Workmentioning

confidence: 99%

Section: Conclusion and Further Workmentioning

confidence: 99%

Section: Conclusion and Further Workmentioning

confidence: 99%

See 1 more Smart Citation

Spatio-Temporal Processing for Automatic Vehicle Detection in Wide-Area Aerial Video

et al. 2020

View full text Add to dashboard Cite

show abstract

“…This shift is primarily attributed to the remarkable capabilities of deep neural networks in handling complex image data. In the subsequent sections [10], [11], it will delve into the current limitations and challenges associated with deep learning-based fruit detection methods, highlighting the need for further research and innovation.…”

Section: Introductionmentioning

confidence: 99%

Detection of Fruit using YOLOv8-based Single Stage Detectors

GAO,

ZHANG

2023

IJACSA

View full text Add to dashboard Cite

In the agricultural sector, the precise detection of fruits plays a pivotal role in optimizing harvesting procedures, minimizing waste, and ensuring the delivery of high-quality produce. Deep learning methods have consistently exhibited superior accuracy compared to alternative techniques, making them a focal point in fruit detection research. However, the ongoing challenge lies in meeting the stringent accuracy requirements essential for real-world applications in agriculture. Addressing this critical concern, this study proposes an innovative solution utilizing the Yolov8 architecture for fruit detection. The methodology involves the meticulous creation of a custom dataset tailored to capture the diverse characteristics of agricultural fruits, followed by rigorous training, validation, and testing processes. Through extensive experimentation and performance evaluations, the findings underscore the exceptional accuracy achieved by the Yolov8-based model. This methodology not only surpasses existing benchmarks but also establishes a robust foundation for transforming fruit detection practices in agriculture. By effectively addressing the challenges associated with accuracy rates, this approach opens new avenues for optimized harvesting, waste reduction, and enhanced efficiency in agricultural practices, contributing significantly to the evolution of precision farming technologies.

show abstract

“…Saliency detection aims at finding the most distinctive objects in an image which are consistent with human visual perception. It is commonly utilized as a preliminary processing step to facilitate a wide range of applications such as object recognition [1], person re-identification [2], image retrieval [3], semantic segmentation [4], scene classification [5], visual tracking [6], video summarization [7] and so on.…”

Section: Introductionmentioning

confidence: 99%

Accurate salient object detection via dense recurrent connections and residual-based hierarchical feature integration

Cao

Yang

et al. 2019

Signal Processing: Image Communication

View full text Add to dashboard Cite

Recently, the convolutional neural network (CNN) has achieved great progress in many computer vision tasks including object detection, image restoration, and scene understanding. In this paper, we propose a novel CNN-based saliency detection method through dense recurrent connections and residual-based hierarchical feature integration. Inspired by the recent neurobiological finding that abundant recurrent connections exist in the human visual system, we firstly propose a novel dense recurrent CNN module (D-RCNN) to learn informative saliency cues by incorporating dense recurrent connections into sub-layers of convolutional stages. Then we present a residual-based architecture with short connections for deep supervision which hierarchically combines both coarse-level and fine-level feature representations. Our end-to-end method takes raw RGB images as input and directly outputs saliency maps without relying on any time-consuming pre/post-processing techniques. Extensive qualitative and quantitative evaluation results on four widely tested benchmark datasets demonstrate that our method can achieve more accurate saliency detection results solutions with significantly fewer model parameters.

show abstract

A parallel spatiotemporal saliency and discriminative online learning method for visual target tracking in aerial videos

Cited by 20 publications

References 47 publications

Spatio-Temporal Processing for Automatic Vehicle Detection in Wide-Area Aerial Video

Spatio-Temporal Processing for Automatic Vehicle Detection in Wide-Area Aerial Video

Detection of Fruit using YOLOv8-based Single Stage Detectors

Accurate salient object detection via dense recurrent connections and residual-based hierarchical feature integration

Contact Info

Product

Resources

About