Visual search difficulty prediction with image ROI information

• Published in: Neural computing & applications Vol. 34; no. 9; pp. 6799 - 6809
• Main Authors: Xiao, Bo, Liu, Xuelian, Wang, Chunyang
• Format: Journal Article
• Language: English
• Published: London Springer London 01.05.2022; Springer Nature B.V
• Subjects: Algorithms; Artificial Intelligence; Computational Biology/Bioinformatics; Computational Science and Engineering; Computer Science; Data Mining and Knowledge Discovery; Image Processing and Computer Vision; Machine learning; Object recognition; Optimization; Prediction models; Probability and Statistics in Computer Science; Regression models; S.I.: NC for Industry 4.0; Searching; Special Issue on Neural Computing Challenges and Applications for Industry 4.0; Support vector machines; Target recognition; Training; Learning using privileged information; Region of interest; Visual search difficulty prediction; Coordinate descent algorithm
• ISSN: 0941-0643; 1433-3058
• DOI: 10.1007/s00521-021-06413-9

Target recognition difficulty quantification and prediction using the search time for the human visual system to target an object is a challenging task, which can effectively guide the training of machine learning models such as target recognition and target location. Our work focuses on how to use region-of-interest (ROI) information to improve the accuracy of the visual search difficulty prediction model. First, the influence of ROI information on visual search difficulty is explored in this paper. Then, based on the learning using privileged information paradigm, we build a support vector regression model using privileged information (SVR +), which uses the deep features of ROIs in the training stage. Next, a coordinate descent algorithm is developed to solve the dual optimization problem in SVR + training. Comprehensive experiments validate the improvement in the accuracy of the proposed model in predicting the difficulty of visual search and the efficiency of our coordinate descent algorithm in model training.