FaultNet: Faulty Rail-Valves Detection using Deep Learning and Computer Vision

Rail track condition monitoring: a review on deep learning approaches

“…Raw data type: it is observed that 70% of studies used image-type raw data for the deep learning models. Nevertheless, acoustic emission signals [65,71,100,103,108] , defectogram [96,109] , speed accelerations [98] , concatenated vector of curve and numbers [101] , current signal [89] , maintenance records [80,99] , synthetic data from generative model [63] , time-frequency measurement data [82] , time-series [60] , geometry data [87] , and vibration signal [119] could all be possible input data sources as well. Purpose of study: it is observed that detection, classification, and/or localizing rail surface defects including various components (rail, insulator, valves, fasteners, switches, track intrusions, etc.)…”

“…Multilayer feedforward neural networks based on multi-valued neurons (MLMVN) [60] neural network [96] Point Cloud deep learning [92] ResNet classifier, DenseNet classifier [81]…”

Machine-Learning Based Methodologies for 3D X-Ray Measurement, Characterization and Optimization for Buried Structures in Advanced IC Packages

“…Recently, Machine learning and Deep learning have made significant strides in improving the performance in object detection and segmentation benchmarks [7,8]. Using novel neural networks, we can perform pixel and sub-pixel level 2D object detection [9] and image segmentation [3]. Significant effort is being focused in extending these techniques to 3D where data is captured using modern tools such as 3D XRMs, LiDAR, and RGBD cameras [10].…”

“…Machine Learning (ML) and Deep Learning (DL) are fast becoming an integral part of advanced manufacturing and inspection fields such as component design [1], optical inspection [2], and anomaly detection [3]. Such systems, with the availability of vast datasets, have improved significantly over classical handcrafted feature learning approaches in various domains.…”

Deep Learning aplicado a inspeção visual da presença de um componente de conjunto de eixo

“…O objetivo da segmentação é particionar uma imagem em regiões com uma aparência visual homogênea razoável, ou que corresponda a objetos ou partes de objetos [28] [39]. Entre as técnicas de segmentação estão: buscar a região de interesse e detectar anomalias usando limiar [37] e redes neurais artificiais como a arquitetura U-Net [40]. A extração de características constitui o cálculo de medidas específicas, que caracterizam o sinal [41] a fim de que as classes possam ser discriminadas.…”