DOI•
Automated and Refined Application of Convolutional Neural Network Modeling to Metallic Powder Particle Satellite Detection
11 Nov 2021-pp 1-16
TL;DR: In this article, the scale-ability of the Mask R-CNN-based model was investigated for the detection and quantification of satellites found within metallic powders, where the original formulated model can be expanded to include scanning electron micrographs to various powder types at variate magnifications.
Abstract: Research concerned with the identification as well as quantification of satellites found within metallic powders has recently demonstrated the promise of implementing Mask R-CNNs, instance segmentation, and transfer learning. Though the original research and development of such an approach demonstrated the functionality of the data-driven image analysis framework, questions remained in regards to the scale-ability of the Mask R-CNN-based model. Accordingly, the present work demonstrates the fact that the originally formulated model can be expanded to include scanning electron micrographs to various powder types at variate magnifications (rather than the original case of micrographs of a single powder type at a single magnification). Moreover, the present work establishes a process that enables users to specifically target which images will have most impact on increasing generalize-ability and performance in order to optimize maximum improvement of the model with the least amount of images annotated. Beyond this, we also outline a method of auto-labeling satellites in images by using a trained model to increase its own training set size.
Citations
More filters
TL;DR: In this paper , the surface evolution of as-produced copper cold-sprayed material consolidations was studied through mass finishing, revealing an inverse correlation relating material removal rate and hardness/strength of the cold sprayed deposits.
Abstract: The surface roughness of additively manufactured (AM) components can have deleterious effects on the properties of the final part, such as corrosion resistance and fatigue life. Modification of the surface finish or parts produced by AM processes, such as cold spray, through methods such as mass finishing, can help to mitigate some of these issues. In this work, the surface evolution of as-produced copper cold sprayed material consolidations was studied through mass finishing. Three different copper powders attained by different production methods and of different sizes were used as feedstock. The surface topography of the cold spray deposits was measured as a function of the mass finishing time for the three copper cold spray samples and analyzed in terms of relative area and complexity, revealing an inverse correlation relating material removal rate and hardness/strength of the cold sprayed deposits. The material removal rate was also affected by the quality of the cold spray deposition, as defined by deposition efficiency (DE). Large initial drops in relative area and complexity are also likely due to the removal of loosely bonded powders at the start of mass finishing. Based on this study, the cold spray parameters that affect the rate of mass finishing have been explored.
4 citations
TL;DR: In this article , the authors evaluate the impact of the model structure and GPU on nondeterminism and identify its exact causes, and propose methods to reduce the amount of variation between model performances while training on a GPU.
Abstract: Convolutional Neural Networks, and many other machine learning algorithms, use Graphical Processing Units (GPUs) instead of Central Processing Units (CPUs) to improve the training time of very large modeling computations. This work evaluates the impact of the model structure and GPU on nondeterminism and identifies its exact causes. The ability to replicate results is quintessential to research, thus nondeterminism must be either removed or significantly reduced. Simple methods are provided so that researchers can: (1) measure the impact of nondeterminism, (2) achieve determinable results by eliminating randomness embedded in the model structure and performing computations on a CPU, or (3) reduce the amount of variation between model performances while training on a GPU.
2 citations
01 Jan 2023
TL;DR: In this article , the authors present a multifaceted consideration of various current cold spray additive deposition advancements, with consideration given primarily to metal materials for cold spray and their counterpart metallurgical material consolidations.
Abstract: This chapter presents a multifaceted consideration of various current cold spray additive deposition advancements. The chapter is divided into five sections, with consideration given primarily to metal materials for cold spray and their counterpart metallurgical material consolidations. The first section considers functional cold-sprayed coatings and applications, including cold-sprayed materials for catalytic and antipathogenic applications. The second considers machine learning, statistical, and data-driven analysis within the cold spray materials literature. The third section examines recent refinements, advancements, and insights obtained concerning bonding mechanisms. After that, the fourth section explores the utility, promise, and potential of thermally preprocessing feedstock before cold spray processing. Finally, the fifth section considers the role of nondestructive evaluation, testing, and analysis for cold spray R&D and quality assurance.
TL;DR: In this paper , an automatic algorithm was proposed to identify spherical powder particles, especially heavily overlapped particles, from their microscope images, and the accuracy and efficiency of the algorithm were validated by real-world scanning electron microscope images.
Abstract: Abstract The microstructural characteristics of spherical metal powders play an important role in determining the quality of mechanical parts manufactured by powder metallurgy processes. Identifying the individual powder particles from their microscopic images is one of the most convenient and cost-efficient methods for the analysis of powder characteristics. Although numerous image processing algorithms have been developed for automating the powder particle identification process, they perform less accurately in identifying adjacent particles that are heavily overlapped in their image regions. We propose an automatic algorithm to robustly and accurately identify spherical powder particles, especially heavily overlapped particles, from their microscope images. A parallel computing framework is designed to further enhance the computational efficiency of the proposed algorithm. Powder characteristics such as particle size distribution and the location of potential satellite particles have been derived from our identification results. The accuracy and efficiency of our algorithm are validated by real-world scanning electron microscope images, outperforming other existing methods and achieving both precision and recall above 99%.
TL;DR: In this article , a data-driven framework based on powder size and shape characteristics for Hall-flow-rate predictions was developed for processing multiple-instance powder data and compared with standard machine learning models.
Abstract: This study investigates the relationship between metallic powders and their flowability behavior (captured in terms of Hall flow rates using Hall flowmeters). Due to the many trait dependencies of powder flowability, which have made the formulation of a physical and mechanistic generalizable model difficult to resolve, this study seeks to develop an alternative data-driven framework based on powder size and shape characteristics for Hall-flow-rate predictions. A multiple-instance regression framework was both developed for processing multiple-instance powder data and compared with standard machine learning models. Data augmentation was found to improve the overall performance of the framework, although the limited dataset was a constraint. Still, the study contributes to ongoing efforts to identify traditional, associative, and generalizable patterns between powder properties and resultant flowability behaviors. The findings show promise for real-world applications with a larger dataset, such that this initial application of multiple instance regression frameworks for metal powder Hall-flow-rate predictions as a function of powder particle size and shape data can be scrutinized in full.
References
More filters
07 Jun 2015
TL;DR: The key insight is to build “fully convolutional” networks that take input of arbitrary size and produce correspondingly-sized output with efficient inference and learning.
Abstract: Convolutional networks are powerful visual models that yield hierarchies of features. We show that convolutional networks by themselves, trained end-to-end, pixels-to-pixels, exceed the state-of-the-art in semantic segmentation. Our key insight is to build “fully convolutional” networks that take input of arbitrary size and produce correspondingly-sized output with efficient inference and learning. We define and detail the space of fully convolutional networks, explain their application to spatially dense prediction tasks, and draw connections to prior models. We adapt contemporary classification networks (AlexNet [20], the VGG net [31], and GoogLeNet [32]) into fully convolutional networks and transfer their learned representations by fine-tuning [3] to the segmentation task. We then define a skip architecture that combines semantic information from a deep, coarse layer with appearance information from a shallow, fine layer to produce accurate and detailed segmentations. Our fully convolutional network achieves state-of-the-art segmentation of PASCAL VOC (20% relative improvement to 62.2% mean IU on 2012), NYUDv2, and SIFT Flow, while inference takes less than one fifth of a second for a typical image.
28,225 citations
TL;DR: Matplotlib is a 2D graphics package used for Python for application development, interactive scripting, and publication-quality image generation across user interfaces and operating systems.
Abstract: Matplotlib is a 2D graphics package used for Python for application development, interactive scripting,and publication-quality image generation across user interfaces and operating systems
23,312 citations
[...]
TL;DR: This work presents a conceptually simple, flexible, and general framework for object instance segmentation, which extends Faster R-CNN by adding a branch for predicting an object mask in parallel with the existing branch for bounding box recognition.
Abstract: We present a conceptually simple, flexible, and general framework for object instance segmentation. Our approach efficiently detects objects in an image while simultaneously generating a high-quality segmentation mask for each instance. The method, called Mask R-CNN, extends Faster R-CNN by adding a branch for predicting an object mask in parallel with the existing branch for bounding box recognition. Mask R-CNN is simple to train and adds only a small overhead to Faster R-CNN, running at 5 fps. Moreover, Mask R-CNN is easy to generalize to other tasks, e.g., allowing us to estimate human poses in the same framework. We show top results in all three tracks of the COCO suite of challenges, including instance segmentation, bounding-box object detection, and person keypoint detection. Without tricks, Mask R-CNN outperforms all existing, single-model entries on every task, including the COCO 2016 challenge winners. We hope our simple and effective approach will serve as a solid baseline and help ease future research in instance-level recognition. Code will be made available.
14,299 citations
Proceedings Article•
07 Dec 2015TL;DR: Ren et al. as discussed by the authors proposed a region proposal network (RPN) that shares full-image convolutional features with the detection network, thus enabling nearly cost-free region proposals.
Abstract: State-of-the-art object detection networks depend on region proposal algorithms to hypothesize object locations. Advances like SPPnet [7] and Fast R-CNN [5] have reduced the running time of these detection networks, exposing region proposal computation as a bottleneck. In this work, we introduce a Region Proposal Network (RPN) that shares full-image convolutional features with the detection network, thus enabling nearly cost-free region proposals. An RPN is a fully-convolutional network that simultaneously predicts object bounds and objectness scores at each position. RPNs are trained end-to-end to generate high-quality region proposals, which are used by Fast R-CNN for detection. With a simple alternating optimization, RPN and Fast R-CNN can be trained to share convolutional features. For the very deep VGG-16 model [19], our detection system has a frame rate of 5fps (including all steps) on a GPU, while achieving state-of-the-art object detection accuracy on PASCAL VOC 2007 (73.2% mAP) and 2012 (70.4% mAP) using 300 proposals per image. Code is available at https://github.com/ShaoqingRen/faster_rcnn.
13,674 citations
Proceedings Article•
01 Jan 2019
TL;DR: This paper details the principles that drove the implementation of PyTorch and how they are reflected in its architecture, and explains how the careful and pragmatic implementation of the key components of its runtime enables them to work together to achieve compelling performance.
Abstract: Deep learning frameworks have often focused on either usability or speed, but not both. PyTorch is a machine learning library that shows that these two goals are in fact compatible: it was designed from first principles to support an imperative and Pythonic programming style that supports code as a model, makes debugging easy and is consistent with other popular scientific computing libraries, while remaining efficient and supporting hardware accelerators such as GPUs. In this paper, we detail the principles that drove the implementation of PyTorch and how they are reflected in its architecture. We emphasize that every aspect of PyTorch is a regular Python program under the full control of its user. We also explain how the careful and pragmatic implementation of the key components of its runtime enables them to work together to achieve compelling performance. We demonstrate the efficiency of individual subsystems, as well as the overall speed of PyTorch on several commonly used benchmarks.
10,045 citations