A novel deep learning approach for tracking with soft computing technique

• Published in: International journal of pervasive computing and communications Vol. 18; no. 5; pp. 678 - 685
• Main Authors: Krishna, Mohan A, Reddy PVN, Satya, Prasad K
• Format: Journal Article
• Language: English
• Published: Bingley Emerald Group Publishing Limited 25.11.2022
• Subjects: Algorithms; Automation; Computer vision; Computers; Deep learning; Feature extraction; Machine learning; Monitoring systems; Optical tracking; Performance assessment; Performance evaluation; Regression; Robotics; Soft computing; Surveillance; Visual discrimination
• ISSN: 1742-7371; 1742-738X
• DOI: 10.1108/IJPCC-08-2021-0196

PurposeIn the community of visual tracking or object tracking, discriminatively learned correlation filter (DCF) has gained more importance. When it comes to speed, DCF gives the best performance. The purpose of this study is to anticipate the object visually. For tracking the object visually, the authors proposed a new model based on the convolutional regression technique. Features like HOG and Harris are used for the process of feature extraction. The authors’ proposed method will give the best results when compared with other existing methods.Design/methodology/approachThe visual tracking of many real-world applications such as robotics, smart monitoring systems, independent driving and human-computer interactions are a major and current research problem in the field of computer vision. This refers to the automated trajectory prediction of an arbitrary target object, often given in the first frame in a bounding box while moving about in successive video frames. In the community of visual tracking or object tracking, DCF has gained more importance. Discriminative trackers strive to train a classifier that differentiates the target item from the background. The fundamental concept is to train a correlation filter that creates high responses around the target and low responses elsewhere. For tracking the object visually, the authors proposed a new model based on the convolutional regression technique. Features like HOG and Harris are used for the process of feature extraction. Through experimental analysis, the authors have evaluated several performance assessment metrics such as accuracy, precision, F-measure and specificity. The authors’ proposed method will give the best results when compared with other existing methods.FindingsThis process involved DCF which gained more importance. When it comes to speed, DCF gives the best performance. The main objective of this study is to anticipate the object visually. For tracking the object visually, the authors proposed a new model based on the convolutional regression technique for tracking the objects and these results will be used for identifying the action of the object.Originality/valueThe main theme exists in the process is to identify the tracking motion of the object by using convolution regression with varied features. This method proves that it will provide better results when compared to state of art methods.