2017 · 29 citations · 10 references
ComputingComputer ScienceComputer VisionDrillingMachine LearningDeep LearningDrilling EquipmentImage ClassificationObject RecognitionPattern RecognitionRepresentation LearningConvolutional Neural NetworksDeep Learning-based ClassificationDrill Bit Wear MonitoringMachine Learning For Drilling
The paper presents an application of transfer learning using convolutional neural network (CNN) in recognition of the drill state on the basis of hole images drilled in the laminated chipboard. Three classes are recognized: red, yellow and green, which correspond with 3 stages of drill state. Red class indicates the drill, which is worn out and should be replaced immediately in drilling process. Yellow class corresponds to the state in which warning should be sent to the operator to check manually state of the drill. The last class corresponds to the green state indicating good condition of drill, enabling further use in production. The important advantage of transfer learning approach is possibility of training classification model using only small portion of data. This is in contrast to the classical deep learning methods of convolutional neural networks, which require very large data base to achieve acceptable accuracy of class recognition. The results of numerical experiments in drill state recognition have confirmed suitability of this novel method to accurate class recognition at small population of available learning data.
10
ImageNet Large Scale Visual Recognition Challenge
Olga Russakovsky, Jia Deng, Hao Su et al. · International Journal of Computer Vision · 2015 · 40.2K citations
Deep Learning: Methods and Applications
Li Deng, Dong Yu · Foundations and Trends® in Signal Processing · 2014 · 3.3K citations · Full text