Data normalization for training and analysis of the MaskRCNN model using the k-means method for computer vision of smart refrigerator

2021 ◽  
Vol 4 ◽  
pp. 74-80
Author(s):  
M. G. Dorrer ◽  
◽  
A.E. Alekhina ◽  
Keyword(s):  
Computer Vision ◽  
Semantic Segmentation ◽  
Correct Choice ◽  
Training Sample ◽  
Artificial Vision ◽  
K Value ◽  
High K ◽  
Real World Datasets ◽  
Segmentation Problem

This paper proposes using the k-means method for the controlled adjustment of the training sample for semantic image segmentation in the artificial vision of a smart refrigerator. To solve this problem, a new two-stage architecture for computer vision is proposed. In the proposed architecture, various sets of settings for optimizing the contrast of images are used to classify pixels according to their belonging to fragments of the studied image. Extensive experimental evaluation shows that the proposed method has critical advantages over existing work. Firstly, the obtained pixel classes can be directly clustered into semantic groups using k-means. Secondly, the method can be used for additional training of artificial intelligence in solving the semantic segmentation problem. The developers propose an approach to the correct choice of the number k of centroids to obtain good quality clusters, which is difficult to determine at a high k value. To overcome the problem of initializing the k-means method, an incremental k-means clustering method is proposed, which improves the quality of clusters to reduce the sum of squared errors. Comprehensive experiments have been carried out compared to the traditional k-means algorithm and its new versions to evaluate the performance of the proposed method on synthetically generated datasets and some real-world datasets.

scholarly journals A COMPARATIVE STUDY OF POINT CLOUDS SEMANTIC SEGMENTATION USING THREE DIFFERENT NEURAL NETWORKS ON THE RAILWAY STATION DATASET

2021 ◽  
Vol XLIII-B3-2021 ◽  
pp. 223-228
Author(s):  
Y. A. Lumban-Gaol ◽  
Z. Chen ◽  
M. Smit ◽  
X. Li ◽  
M. A. Erbaşu ◽  
...  
Keyword(s):  
Neural Networks ◽  
Point Cloud ◽  
Semantic Segmentation ◽  
Point Clouds ◽  
Railway Station ◽  
Train Station ◽  
Cloud Data ◽  
Indoor Scenes ◽  
Segmentation Problem

Abstract. Point cloud data have rich semantic representations and can benefit various applications towards a digital twin. However, they are unordered and anisotropically distributed, thus being unsuitable for a typical Convolutional Neural Networks (CNN) to handle. With the advance of deep learning, several neural networks claim to have solved the point cloud semantic segmentation problem. This paper evaluates three different neural networks for semantic segmentation of point clouds, namely PointNet++, PointCNN and DGCNN. A public indoor scene of the Amersfoort railway station is used as the study area. Unlike the typical indoor scenes and even more from the ubiquitous outdoor ones in currently available datasets, the station consists of objects such as the entrance gates, ticket machines, couches, and garbage cans. For the experiment, we use subsets from the data, remove the noise, evaluate the performance of the selected neural networks. The results indicate an overall accuracy of more than 90% for all the networks but vary in terms of mean class accuracy and mean Intersection over Union (IoU). The misclassification mainly occurs in the classes of couch and garbage can. Several factors that may contribute to the errors are analyzed, such as the quality of the data and the proportion of the number of points per class. The adaptability of the networks is also heavily dependent on the training location: the overall characteristics of the train station make a trained network for one location less suitable for another.

Scalability of MOCVD-deposited Hafnium Oxide

2003 ◽  
Vol 765 ◽  
Author(s):  
S. Van Elshocht ◽  
R. Carter ◽  
M. Caymax ◽  
M. Claes ◽  
T. Conard ◽  
...  
Keyword(s):  
Hafnium Oxide ◽  
Interfacial Layer ◽  
Gate Dielectric ◽  
Deposition Process ◽  
Oxide Thickness ◽  
Process Conditions ◽  
Primary Concern ◽  
Gate Electrode ◽  
K Value ◽  
High K

AbstractBecause of aggressive downscaling to increase transistor performance, the physical thickness of the SiO2 gate dielectric is rapidly approaching the limit where it will only consist of a few atomic layers. As a consequence, this will result in very high leakage currents due to direct tunneling. To allow further scaling, materials with a k-value higher than SiO2 (“high-k materials”) are explored, such that the thickness of the dielectric can be increased without degrading performance.Based on our experimental results, we discuss the potential of MOCVD-deposited HfO2 to scale to (sub)-1-nm EOTs (Equivalent Oxide Thickness). A primary concern is the interfacial layer that is formed between the Si and the HfO2, during the MOCVD deposition process, for both H-passivated and SiO2-like starting surfaces. This interfacial layer will, because of its lower k-value, significantly contribute to the EOT and reduce the benefit of the high-k material. In addition, we have experienced serious issues integrating HfO2 with a polySi gate electrode at the top interface depending on the process conditions of polySi deposition and activation anneal used. Furthermore, we have determined, based on a thickness series, the k-value for HfO2 deposited at various temperatures and found that the k-value of the HfO2 depends upon the gate electrode deposited on top (polySi or TiN).Based on our observations, the combination of MOCVD HfO2 with a polySi gate electrode will not be able to scale below the 1-nm EOT marker. The use of a metal gate however, does show promise to scale down to very low EOT values.

scholarly journals Clay-Based Brick Porosity Estimation Using Image Processing Techniques

2019 ◽  
Vol 29 (1) ◽  
pp. 1226-1234
Author(s):  
Safa Jida ◽  
Hassan Ouallal ◽  
Brahim Aksasse ◽  
Mohammed Ouanan ◽  
Mohamed El Amraoui ◽  
...  
Keyword(s):  
Electron Microscopy ◽  
Image Processing ◽  
Scanning Electron Microscopy ◽  
Computer Vision ◽  
Image Processing Techniques ◽  
Processing Techniques ◽  
Porosity Estimation ◽  
Microscopy Images ◽  
Scanning Electron

Abstract This work intends to apprehend and emphasize the contribution of image-processing techniques and computer vision in the treatment of clay-based material known in Meknes region. One of the various characteristics used to describe clay in a qualitative manner is porosity, as it is considered one of the properties that with “kill or cure” effectiveness. For this purpose, we use scanning electron microscopy images, as they are considered the most powerful tool for characterising the quality of the microscopic pore structure of porous materials. We present various existing methods of segmentation, as we are interested only in pore regions. The results show good matching between physical estimation and Voronoi diagram-based porosity estimation.

Vision-Based Precision Deburring and Edge Breaking Process Quality Control

2020 ◽  
Author(s):  
Kartik Gupta ◽  
Cindy Grimm ◽  
Burak Sencer ◽  
Ravi Balasubramanian
Keyword(s):  
Computer Vision ◽  
Quality Control ◽  
Vision System ◽  
Quality Measures ◽  
Cad Model ◽  
Computer Vision System ◽  
Part Geometry ◽  
Specialized Hardware ◽  
Original Part

Abstract This paper presents a computer vision system for evaluating the quality of deburring and edge breaking on aluminum and steel blocks. This technique produces both quantitative (size) and qualitative (quality) measures of chamfering operation from images taken with an off-the-shelf camera. We demonstrate that the proposed computer vision system can detect edge chamfering geometry within a 1–2mm range. The proposed technique does not require precise calibration of the camera to the part nor specialized hardware beyond a macro lens. Off-the-shelf components and a CAD model of the original part geometry are used for calibration. We also demonstrate the effectiveness of the proposed technique on edge breaking quality control.

scholarly journals Nb2O5 and Ti-Doped Nb2O5 Charge Trapping Nano-Layers Applied in Flash Memory

Nanomaterials ◽  
2018 ◽  
Vol 8 (10) ◽  
pp. 799 ◽  
Author(s):  
Jer Wang ◽  
Chyuan Kao ◽  
Chien Wu ◽  
Chun Lin ◽  
Chih Lin
Keyword(s):  
Flash Memory ◽  
Charge Trapping ◽  
Memory Device ◽  
Dielectric Constants ◽  
Oxide Semiconductor ◽  
Flat Band ◽  
Hysteresis Curve ◽  
Band Shift ◽  
K Value ◽  
High K

High-k material charge trapping nano-layers in flash memory applications have faster program/erase speeds and better data retention because of larger conduction band offsets and higher dielectric constants. In addition, Ti-doped high-k materials can improve memory device performance, such as leakage current reduction, k-value enhancement, and breakdown voltage increase. In this study, the structural and electrical properties of different annealing temperatures on the Nb2O5 and Ti-doped Nb2O5(TiNb2O7) materials used as charge-trapping nano-layers in metal-oxide-high k-oxide-semiconductor (MOHOS)-type memory were investigated using X-ray diffraction (XRD) and atomic force microscopy (AFM). Analysis of the C-V hysteresis curve shows that the flat-band shift (∆VFB) window of the TiNb2O7 charge-trapping nano-layer in a memory device can reach as high as 6.06 V. The larger memory window of the TiNb2O7 nano-layer is because of a better electrical and structural performance, compared to the Nb2O5 nano-layer.

scholarly journals Hydrogen Diffusion Coefficient, Hydrogen Solution Coefficient and Hydrogen Permeability of Nb-TiNi Eutectic Alloy

2010 ◽  
Vol 638-642 ◽  
pp. 1131-1136
Author(s):  
Wei Liang Wang ◽  
Kazuhiro Ishikawa ◽  
Kiyoshi Aoki
Keyword(s):  
Diffusion Coefficient ◽  
Hydrogen Absorption ◽  
Hydrogen Diffusion ◽  
Hydrogen Permeability ◽  
K Value ◽  
Hydrogen Flow ◽  
Hydrogen Diffusion Coefficient ◽  
High K ◽  
Pressure Region ◽  
The Difference

In general, hydrogen permeabilityΦ of the alloy membrane is expressed as the product of the hydrogen diffusion coefficient D and the hydrogen solution coefficient K. Therefore, to improve the hydrogen permeability efficiently, the values of K and D should be separately considered. In the present study, hydrogen absorption and permeation behaviors of the Nb19Ti40Ni41 alloy consisting of the eutectic phase are investigated by measuring pressure-composition-isotherm (PCI) and by the hydrogen flow method and compared with those of palladium. The hydrogen absorption in the Nb19Ti40Ni41 alloy does not obey the Sieverts’ law in the pressure region of 0-1.0MPa at 523K, but it shows linear relationship between the difference in the square root of hydrogen pressure and hydrogen content between 0.1 and 0.4MPa. Although the value of D for the Nb19Ti40Ni41 alloy is considerably lower than that of palladium, its high K value enhances the hydrogen permeability Φ. It is suggested that the enhancement of D by microstructural control for Nb19Ti40Ni41 alloy is effective for improvement of Φ.

Physical-Chemical Evolution of Hf-aluminates upon Thermal Treatments

2003 ◽  
Vol 765 ◽  
Author(s):  
B. Crivelli ◽  
M. Alessandri ◽  
S. Alberici ◽  
D. Brazzelli ◽  
A. C. Elbaz ◽  
...  
Keyword(s):  
Orthorhombic Phase ◽  
Interface Layer ◽  
Film Structure ◽  
Thermal Treatments ◽  
Tem Analysis ◽  
K Value ◽  
Physical Chemical ◽  
High K ◽  
Post Deposition ◽  
Limited Oxygen

AbstractThis study presents an investigation on physical-chemical stability of (HfO2)x(Al2O3 )1-x alloys upon prolonged post-deposition annealings. Two different Hf-aluminates were deposited by ALCVDTM, containing 34% and 74% Al2O3 mol% respectively. Post-deposition annealings (PDA) were carried out in O2 or N2 atmosphere, at 850°C and 900°C for 30 minutes. Interfacial layer (IL) increase after PDA was detected on all the samples, but with small differences between N2 and O2 treatments. Stack composition was characterized by means of XRR, XRF, RBS and TOF-SIMS. Growth of interface layer was justified by limited oxygen incorporation from external ambient. Silicon diffusion from the substrate into high-k material and aluminum/hafnium redistribution were observed and associated to annealing temperature. XRD and planar TEM analysis evidenced first grain formation and then, in the case of Hf-rich samples, almost complete crystallization. Overall, Hf-aluminates were found to remain XRD amorphous during high temperature prolonged treatments up to 900°C for 74% and 850°C for 34% alloys respectively. Differently from HfO2, (HfO2)0.66(Al2O3 )0.34 alloy was observed to crystallized in orthorhombic phase. Hf-aluminates were also electrically characterized by means of C(V) and I(V) measurements on basic capacitors. Variations in material electrical properties were found consistent with change in physical-chemical film structure. Increase in k value up to 30 was observed on Hf-rich samples crystallized in orthorhombic phase.

scholarly journals EVALUATION OF SEMANTIC SEGMENTATION METHODS FOR DEFORESTATION DETECTION IN THE AMAZON

2020 ◽  
Vol XLIII-B3-2020 ◽  
pp. 1497-1505
Author(s):  
R. B. Andrade ◽  
G. A. O. P. Costa ◽  
G. L. A. Mota ◽  
M. X. Ortega ◽  
R. Q. Feitosa ◽  
...  
Keyword(s):  
Global Climate ◽  
Biodiversity Loss ◽  
Semantic Segmentation ◽  
Training Sample ◽  
Training Data ◽  
Detection Methods ◽  
Amazon Forest ◽  
Convolutional Network ◽  
Detection Approach ◽  
Segmentation Methods

Abstract. Deforestation is a wide-reaching problem, responsible for serious environmental issues, such as biodiversity loss and global climate change. Containing approximately ten percent of all biomass on the planet and home to one tenth of the known species, the Amazon biome has faced important deforestation pressure in the last decades. Devising efficient deforestation detection methods is, therefore, key to combat illegal deforestation and to aid in the conception of public policies directed to promote sustainable development in the Amazon. In this work, we implement and evaluate a deforestation detection approach which is based on a Fully Convolutional, Deep Learning (DL) model: the DeepLabv3+. We compare the results obtained with the devised approach to those obtained with previously proposed DL-based methods (Early Fusion and Siamese Convolutional Network) using Landsat OLI-8 images acquired at different dates, covering a region of the Amazon forest. In order to evaluate the sensitivity of the methods to the amount of training data, we also evaluate them using varying training sample set sizes. The results show that all tested variants of the proposed method significantly outperform the other DL-based methods in terms of overall accuracy and F1-score. The gains in performance were even more substantial when limited amounts of samples were used in training the evaluated methods.

scholarly journals Automatic Instrument Segmentation in Robot-Assisted Surgery Using Deep Learning

2018 ◽  
Author(s):  
Alexey A. Shvets ◽  
Alexander Rakhlin ◽  
Alexandr A. Kalinin ◽  
Vladimir I. Iglovikov
Keyword(s):  
Deep Learning ◽  
State Of The Art ◽  
Semantic Segmentation ◽  
Robot Assisted ◽  
Automatic Instrument ◽  
Binary Segmentation ◽  
Video Feed ◽  
Class Segmentation ◽  
Segmentation Problem ◽  
Robot Assisted Surgery

AbstractSemantic segmentation of robotic instruments is an important problem for the robot-assisted surgery. One of the main challenges is to correctly detect an instrument’s position for the tracking and pose estimation in the vicinity of surgical scenes. Accurate pixel-wise instrument segmentation is needed to address this challenge. In this paper we describe our deep learning-based approach for robotic instrument segmentation. Our approach demonstrates an improvement over the state-of-the-art results using several novel deep neural network architectures. It addressed the binary segmentation problem, where every pixel in an image is labeled as an instrument or background from the surgery video feed. In addition, we solve a multi-class segmentation problem, in which we distinguish between different instruments or different parts of an instrument from the background. In this setting, our approach outperforms other methods for automatic instrument segmentation thereby providing state-of-the-art results for these problems. The source code for our solution is made publicly available.

Sign in / Sign up

Export Citation Format

Share Document