Authors

Neural network approach for detection of defects “weft crack” and “water damage” in textile fabrics

At present, the automation of production processes, including the use of computer vision, machine learning and artificial intelligence methods, is of relevance at light industry enterprises due to the fourth industrial revolution. The key role in the production processes is played by the quality of manufactured products – textile fabrics, which is directly affected by the process of defectoscopy. Due to the development of digital technologies and the growth of computing power, it is possible to automate the process of defectoscopy of textile fabrics using computer vision to reduce labor costs and increase the accuracy of defect detection. The purpose of this paper is to conduct experimental studies of the marking and detection of specific classes of textile defects using a hardware-software complex of computer vision and using a neural network approach. To achieve this goal, the paper describes the existing classification of textile web defects, describes the used hardware-software system, and presents the application of the neural network model of the Mask R-CNN architecture to solve the problem of exemplar defect segmentation. As part of the study, a manual partitioning of more than 400 tissue photographs into two classes of defects was performed as an extension of the training sample: “weft crack” and “water damage”, the obtained results of the neural network model were evaluated by IoU metrics: the best result for the class “weft crack” DIoU = 0.2, for the class “water damage” DIoU = 0.87. Based on the results of the experimental studies, conclusions are made about the existing potential of using neural network approach for defectoscopy of similar classes of defects. The presented results can be used for training and retraining of various models of object detection, the gained experience can be applied in other spheres of industry. Read more...

A generalized attention model for survival analysis of complex objects under censored data

A wide variety of applied fields, including medicine, security, economics, and industry, are concerned with modeling the processes of various events occurring, such as a patient’s recovery, a company’s financial bankruptcy, industrial equipment failure, etc. Their modeling can be performed within the framework of survival analysis, a statistical method for analyzing time-to-event data whose distinctive feature, setting it apart from many other statistical and machine learning methods, is the presence of censored data. This occurs when an event is not observed and it is only known that it did not happen before a certain point in time. Censored data significantly complicates the modeling and prediction of critical events. Machine learning is an effective tool for survival analysis in the presence of censored data. In particular, modern transformer-based machine learning models demonstrate promising results in survival analysis due to their ability to account for complex dependencies. However, the standard attention mechanism in these models often ignores the fundamental structure of time-to-event data, namely, the distinction between censored and uncensored observations. To overcome this shortcoming, this paper proposes a new model and a new approach to implementing an attention mechanism that redefines attention weights by incorporating prior characteristics of survival analysis based on the Beran estimator or the Cox model. Instead of relying solely on distances between feature vector representations, as is done in current models, the proposed model computes attention weights as a weighted linear combination of components derived from key prior characteristics of survival analysis, such as distances between survival function estimates or time-to-event expectations for different training objects. The proposed approach enables a significant expansion of the class of transform models for survival analysis, achieving higher prediction accuracy. The algorithm implementing the proposed model is the basis for transformers. Experiments on real datasets confirm that the generalized model provides the best prediction among a number of known models. Read more...