Answers are grounded in this paper and cite their sources.
Knot image classification includes the usage of computer vision technology, along with the Machine Learning (ML) method, to classify and recognize various types of knots from images. The main task of knot theory is to authorize a knot invariant, which could be properly differentiated between diverse knot types. Therefore, this work introduced an advanced hybrid deep learning model for knot image classification. Initially, knot images are given to Region of Interest (ROI) extraction using color-based thresholding. Then, the process of feature extraction is performed, where the extraction and identification of the important features from the raw data are done to create a very informative dataset. Thus, Hessian-based Laplacian of Gaussian (HLOG), Shape Local Binary Texture (SLBT), and Hierarchical skeleton feature, along with Convolutional Neural Network (CNN) features, are used for the feature extraction. Lastly, for Knot classification, proposed Neural Search Architecture Network-Residual Neural Network 152 (NasNet-ResNet152) is employed, which is created by hybridizing ResNet and NasNet models. At last, the experimental analysis is conducted regarding metrics, such as sensitivity, accuracy, and specificity. The results state that the NasNet-ResNet152 method reached an accuracy of 96.4%, sensitivity of 96.6%, and specificity of 96.3%.
No takes yet. Share an insight, caveat, or question.
Chen et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: