Now showing 1 - 10 of 14
No Thumbnail Available
Publication

Click-event sound detection in automotive industry using machine/deep learning

2021 , Espinosa Loera, Ricardo Abel , Ponce, Hiram , Gutiérrez, Sebastián

In the automotive industry, despite the robotic systems on the production lines, factories continue employing workers in several custom tasks getting for semi-automatic assembly operations. Specifically, the assembly of electrical harnesses of engines comprises a set of connections between electrical components. Despite the task is easy to perform, employees tend not to notice that a few components are not being connected properly due to physical fatigue provoked by repetitive tasks. This yields a low quality of the assembly production line and possible hazards. In this work, we propose a sound detection system based on machine/deep learning (ML/DL) approaches to identify click sounds produced when electrical harnesses are connected. The purpose of this system is to count the number of connections properly made and to feedback to the employees. We collect and release a public dataset of 25,000 click sounds of 25 ms length at 22 kHz during three months of assembly operations in an automotive production line located in Mexico. Then, we design an ML/DL-based methodology for click sound detection of assembled harnesses under real conditions of a noisy environment (noise level ranging from −16.67 dB to −12.87 dB) including other machinery sounds. Our best ML/DL model (i.e., a combination between five acoustic features and an optimized convolutional neural network) is able to detect click sounds in a real assembly production line with an accuracy of 94.55±0.83 %. To the best of our knowledge, this is the first time a click sounds detection system in assembling electrical harnesses of engines for giving feedback to the workers is proposed and implemented in a real-world automotive production line. We consider this work valuable for the automotive industry on how to apply ML/DL approaches for improving the quality of semi-automatic assembly operations. © 2021 Elsevier B.V.

Thumbnail Image
Publication

Un enfoque basado en la visión para la detección de caídas utilizando múltiples cámaras y redes neuronales convolucionales: un caso de estudio en UP-Fall Detection Data-Set

2019-12 , Espinosa Loera, Ricardo Abel , HIRAM EREDIN PONCE ESPINOSA;376768 , JOSÉ SEBASTIÁN GUTIÉRREZ CALDERÓN;494470 , Ponce, Hiram , Gutiérrez, Sebastián , Campus Aguascalientes

Actualmente, el reconocimiento automático de caídas humanas es un tema de investigación importante para la visión por computadora y la comunidad de la inteligencia artificial. Para el análisis de imágenes, es común usar un enfoque basado en visión para la detección de caídas y sistemas de clasificación debido al aumento exponencial actual en el uso de cámaras. Además, las técnicas de deep learning han revolucionado las técnicas basadas en visión. Han sido consideradas robustas y confiables en la detección y clasificación de problemas, principalmente usando Redes Neuronales Convolucionales (CNN). Recientemente, nuestro grupo de investigación lanzo un nuevo Data Set multimodal para la detección de caídas (Up-Fall Detecction dataset), y se requieren diferentes estudios de enfoques de modalidades para la detección y clasificación de caídas. Centrándonos solo en un enfoque basado en visión, en este articulo presentamos un sistema de detección de caídas basado en 2D CNN como método de inferencia y varias cámaras. Este enfoque analiza imágenes en marcos de ventana de tiempo fijo que extraen características utilizando un método de flujo óptico que obtiene información de movimiento relativo entre dos imágenes consecutivas. Para resultados experimentales, probamos este enfoque en nuestro dataset público. Los resultados mostraron que nuestra propuesta de enfoque basado en la visión múltiple detecta caídas humanas que alcanzan un 95.64% de precisión con una arquitectura de red CNN simple en comparación con otros métodos de vanguardia.

No Thumbnail Available
Publication

Intelligent Management System for Micro-Grids using Internet-of-Things

2021 , Gutiérrez, Sebastián , Medina, Guillermo , Ponce, Hiram , Espinosa Loera, Ricardo Abel

The current research work proposes the design of an intelligent platform for managing the generation, distribution, transmission, commercialization and consumption of electricity, allowing the decision-making process aimed at reducing costs and maximizing the use of energy resources. The technological proposal is a system of information made of integrated sensors. This creates an electrical real-time network that share energy in the cloud through LPWAN networks. Later, the data from the sensors are received in an intelligent platform (Max4 IoT) that employs super-computing systems and artificial intelligence for the analysis of individual and aggregated data. The system is able to learn about the electrical network, knowing the consumers’ behavior and energy trends, allowing to generate responses based on specific situations by sending SMS alerts and emails about abnormal situations or programmed tasks. In addition, the system allows the generation of reports on the network status in real-time and commands control signals on the switching on-and-off of equipment in response to consumption peaks and/or power supply outages. The platform allows interaction with distributed generation systems and micro-grids, through its mobile or web interface, increasing user interaction with the electrical system and the inclusion of renewable energy sources, thus influencing a better quality of service provided to the end user. © 2021 IEEE.

No Thumbnail Available
Publication

Click Event Sound Detection Using Machine Learning in Automotive Industry

2020 , Espinosa Loera, Ricardo Abel , Ponce, Hiram , Gutiérrez, Sebastián , Hernández Cornu, Javier Eluney

Artificial intelligence has been playing an important role when it comes to the automotive industry and its quality of assemblies in the production line, this is because since the arrival of the industry 4.0 it has been subject to change and continuous improvement. In the past, we’ve observed how many machine learning architectures have been used to create environmental sound classification systems in order to improve traditional systems, thus overcoming efficiency issues with great results. In this work, we present a machine learning solution/approach for click event sound detection using audio sensors that are used in the assembly of electric harnesses for engines, this being done on an automotive production line, where we divided our workflow into: data collection, pre-processing, feature extraction, training and inference and finally the detection of the click event sounds. We created a dataset that is composed by 25,000 audio files that have an average duration of 0.025 seconds per click sound with the purpose of training a Multi-layer Perceptron and bring it into the inference phase. In order to test this approach, we’ve performed various implementations in a laboratory and in the real automotive industry. We obtained 95.23% in F1-Score Metric in a laboratory, while in real conditions, we obtained less reliable results, as 84.00% as the best results. © 2020, Springer Nature Switzerland AG.

No Thumbnail Available
Publication

Renewable Energy Prediction through Machine Learning Algorithms

2020 , Luisa Fernanda Jimenez Alvarez , Sebastian Ramos Gonzalez , Antonio Delgado Lopez , Diego Alonso Hernandez Delgado , Espinosa Loera, Ricardo Abel , Gutiérrez, Sebastián

No Thumbnail Available
Publication

A vision-based approach for fall detection using multiple cameras and convolutional neural networks: A case study using the UP-Fall detection dataset

2019 , Espinosa Loera, Ricardo Abel , Ponce, Hiram , Moya-Albor, Ernesto , Martinez-Villaseñor, Lourdes , Brieva, Jorge , Gutiérrez, Sebastián

The automatic recognition of human falls is currently an important topic of research for the computer vision and artificial intelligence communities. In image analysis, it is common to use a vision-based approach for fall detection and classification systems due to the recent exponential increase in the use of cameras. Moreover, deep learning techniques have revolutionized vision-based approaches. These techniques are considered robust and reliable solutions for detection and classification problems, mostly using convolutional neural networks (CNNs). Recently, our research group released a public multimodal dataset for fall detection called the UP-Fall Detection dataset, and studies on modality approaches for fall detection and classification are required. Focusing only on a vision-based approach, in this paper, we present a fall detection system based on a 2D CNN inference method and multiple cameras. This approach analyzes images in fixed time windows and extracts features using an optical flow method that obtains information on the relative motion between two consecutive images. We tested this approach on our public dataset, and the results showed that our proposed multi-vision-based approach detects human falls and achieves an accuracy of 95.64% compared to state-of-the-art methods with a simple CNN network architecture. © 2019 Elsevier Ltd

No Thumbnail Available
Publication

A Novel Hybrid Endoscopic Dataset for Evaluating Machine Learning-Based Photometric Image Enhancement Models

2022 , Axel García-Vega , Ricardo Espinosa , Gilberto Ochoa-Ruiz , Thomas Bazin , Luis Falcón-Morales , Dominique Lamarque , Christian Daul

Endoscopy is the most widely used medical technique for cancer and polyp detection inside hollow organs. However, images acquired by an endoscope are frequently affected by illumination artefacts due to the enlightenment source orientation. There exist two major issues when the endoscope’s light source pose suddenly changes: overexposed and underexposed tissue areas are produced. These two scenarios can result in misdiagnosis due to the lack of information in the affected zones or hamper the performance of various computer vision methods (e.g., SLAM, structure from motion, optical flow) used during the non invasive examination. The aim of this work is two-fold: i) to introduce a new synthetically generated data-set generated by a generative adversarial techniques and ii) and to explore both shallow based and deep learning-based image-enhancement methods in overexposed and underexposed lighting conditions. Best quantitative results (i.e., metric based results), were obtained by the deep learning-based LMSPEC method, besides a running time around 7.6 fps.

No Thumbnail Available
Publication

An Intelligent Water Consumption Prediction System based on Internet of Things

2020 , Gutiérrez, Sebastián , Ponce, Hiram , Espinosa Loera, Ricardo Abel

This work presents the development of a measurement system for water consumption based on the Internet of Things concept. In this paper, we propose a supervised learning method namely artificial hydrocarbon networks (AHN) to predict water consumption one hour ahead. A Hall effect sensor was used to obtain the water flow value through an embedded system and to show it in an interface developed in Visual Studio. For that, the embedded system sent the data in real time to a database in Firebase using the JSON communication protocol. There, the consumed water flow is stored periodically. Experimental results of the supervised learning model conclude that AHN model predicts the conditions for efficient consumption with an average root-mean squared error of 2.4924 liters per hour. © 2020 IEEE.

No Thumbnail Available
Publication

Multi-Scale Structural-aware Exposure Correction for Endoscopic Imaging

2023 , Axel García-Vega , Espinosa Loera, Ricardo Abel , Luis Ramírez-Guzmán , Thomas Bazin , Luis Falcón-Morales , Gilberto Ochoa-Ruiz , Dominique Lamarque , Christian Daul

No Thumbnail Available
Publication

A 3D orthogonal vision-based band-gap prediction using deep learning: A proof of concept

2022 , Espinosa Loera, Ricardo Abel , Ponce, Hiram , ORTIZ-MEDINA, JOSUE

In this work, a vision-based system for the electronic band-gap prediction of organic molecules is proposed using a multichannel 2D convolutional neural network (CNN) and a 3D CNN, applied to the recognition and classification of 2D projected images from 3D molecular structure models. The generated images are input into the CNN for an estimation of the energy gap, associated with the molecular structure. The public data set used in this research was the Organic Materials Database (OMDB-GAP1). A data transformation from the descriptive information contained in the data set to three 2D orthogonal images of molecules was done. The training set is composed of 30,000 images, whereas the testing set was composed of 7500 images, from 12,500 different molecules. The multichannel 2D CNN architecture was optimized via Bayesian optimization. Experimental results showed that the proposed CNN model obtained an acceptable mean absolute error of 0.6780 eV and root mean-squared error of 0.7673 eV, in contrast to two machine learning methods reported in the literature used for band-gap prediction based on conventional density function theory (DFT) methods. These results demonstrate the feasibility of CNN models to materials science routines using orthogonal images projections of molecules. © 2021 Elsevier B.V.