scholarly journals USING NEURAL NETWORKS TO DETECT OBJECTS IN MLS POINT CLOUDS BASED ON LOCAL POINT NEIGHBORHOODS

2019 ◽  
Vol IV-2/W7 ◽  
pp. 17-24
Author(s):  
B. Borgmann ◽  
M. Hebel ◽  
M. Arens ◽  
U. Stilla
Keyword(s):  
Neural Network ◽  
Point Clouds ◽  
Training Data ◽  
Actual Object ◽  
Object A ◽  
The Neural Network ◽  
Local Point ◽  
Object Center ◽  
Training Examples ◽  
Multiple Neighborhoods

<p><strong>Abstract.</strong> This paper presents an approach which uses a <i>PointNet</i>-like neural network to detect objects of certain types in MLS point clouds. In our case, it is used for the detection of pedestrians, but the approach can easily be adapted to other object classes. In the first step, we process local point neighborhoods with the neural network to determine a descriptive feature. This is then further processed to generate two outputs of the network. The first output classifies the neighborhood and determines if it is part of an object of interest. If this is the case, the second output determines where it is located in relation to the object center. This regression output allows us to use a voting process for the actual object detection. This processing step is inspired by approaches based on implicit shape models (ISM). It is able to deal with a certain amount of incorrectly classified neighborhoods, since it combines the results of multiple neighborhoods for the detection of an object. A benefit of our approach as compared to other machine learning methods is its low demand for training data. In our experiments, we achieved a promising detection performance even with less than 1000 training examples.</p>

scholarly journals PEDESTRIAN DETECTION AND TRACKING IN SPARSE MLS POINT CLOUDS USING A NEURAL NETWORK AND VOTING-BASED APPROACH

2020 ◽  
Vol V-2-2020 ◽  
pp. 187-194
Author(s):  
B. Borgmann ◽  
M. Hebel ◽  
M. Arens ◽  
U. Stilla
Keyword(s):  
Neural Network ◽  
Pedestrian Detection ◽  
Ground Plane ◽  
Point Clouds ◽  
3D Tracking ◽  
Actual Object ◽  
Detection And Tracking ◽  
The Neural Network ◽  
Local Point ◽  
Point Density

Abstract. This paper presents and extends an approach for the detection of pedestrians in unstructured point clouds resulting from single MLS (mobile laser scanning) scans. The approach is based on a neural network and a subsequent voting process. The neural network processes point clouds subdivided into local point neighborhoods. The member points of these neighborhoods are directly processed by the network, hence a conversion in a structured representation of the data is not needed. The network also uses meta information of the neighborhoods themselves to improve the results, like their distance to the ground plane. It decides if the neighborhood is part of an object of interest and estimates the center of said object. This information is then used in a voting process. By searching for maxima in the voting space, the discrimination between an actual object and incorrectly classified neighborhoods is made. Since a single labeled object can be subdivided into multiple local neighborhoods, we are able to train the neural network with comparatively low amounts of labeled data. Considerations are made to deal with the varying and sparse point density that is typical for single MLS scans. We supplement the detection with a 3D tracking which, although straightforward, allows us to deal with objects which are occluded for short periods of time to improve the quality of the results. Overall, our approach performs reasonably well for the detection and tracking of pedestrians in single MLS scans as long as the local point density is not too low. Given the LiDAR sensor we used, this is the case up to distances of 22 m.

scholarly journals CLASSIFICATION OF AERIAL LASER SCANNING POINT CLOUDS USING MACHINE LEARNING: A COMPARISON BETWEEN RANDOM FOREST AND TENSORFLOW

2019 ◽  
Vol XLII-2/W13 ◽  
pp. 1105-1111 ◽  
Author(s):  
F. Pirotti ◽  
F. Tonion
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Random Forest ◽  
Laser Scanning ◽  
Laser Scanner ◽  
Point Clouds ◽  
Training Data ◽  
Semantic Classification ◽  
The Neural Network

<p><strong>Abstract.</strong> In this investigation a comparison between two machine learning (ML) models for semantic classification of an aerial laser scanner point cloud is presented. One model is Random Forest (RF), the other is a multi-layer neural network, TensorFlow (TF). Accuracy results were compared over a growing set of training data, using a stratified independent sampling over classes from 5% to 50% of the total dataset. Results show RF to have average F1&amp;thinsp;=&amp;thinsp;0.823 for the 9 classes considered, whereas TF had average F1&amp;thinsp;=&amp;thinsp;0.450. F1 values where higher for RF than TF, due to complexity in the determination of a suitable composition of the hidden layers of the neural network in TF, and this can likely be improved to reach higher accuracy values. Further study in this sense is planned.</p>

scholarly journals Rethinking the Random Cropping Data Augmentation Method Used in the Training of CNN-Based SAR Image Ship Detector

2020 ◽  
Vol 13 (1) ◽  
pp. 34
Author(s):  
Rong Yang ◽  
Robert Wang ◽  
Yunkai Deng ◽  
Xiaoxue Jia ◽  
Heng Zhang
Keyword(s):  
Neural Network ◽  
Data Augmentation ◽  
Back Propagation ◽  
Detection Performance ◽  
Training Data ◽  
Sar Image ◽  
Optical Images ◽  
The Neural Network ◽  
Effective Training ◽  
Standard Configuration

The random cropping data augmentation method is widely used to train convolutional neural network (CNN)-based target detectors to detect targets in optical images (e.g., COCO datasets). It can expand the scale of the dataset dozens of times while consuming only a small amount of calculations when training the neural network detector. In addition, random cropping can also greatly enhance the spatial robustness of the model, because it can make the same target appear in different positions of the sample image. Nowadays, random cropping and random flipping have become the standard configuration for those tasks with limited training data, which makes it natural to introduce them into the training of CNN-based synthetic aperture radar (SAR) image ship detectors. However, in this paper, we show that the introduction of traditional random cropping methods directly in the training of the CNN-based SAR image ship detector may generate a lot of noise in the gradient during back propagation, which hurts the detection performance. In order to eliminate the noise in the training gradient, a simple and effective training method based on feature map mask is proposed. Experiments prove that the proposed method can effectively eliminate the gradient noise introduced by random cropping and significantly improve the detection performance under a variety of evaluation indicators without increasing inference cost.

Application of ANN Back-Propagation for Fracture Design Parameters in Extra-Low Cycle Rotating Bending Fatigue

2009 ◽  
Vol 610-613 ◽  
pp. 450-453
Author(s):  
Hong Yan Duan ◽  
You Tang Li ◽  
Jin Zhang ◽  
Gui Ping He
Keyword(s):  
Neural Network ◽  
Neural Network Model ◽  
Back Propagation ◽  
Training Data ◽  
Design Parameters ◽  
Ann Model ◽  
Bending Fatigue ◽  
The Neural Network ◽  
Rotating Bending Fatigue ◽  
Rotating Bending

The fracture problems of ecomaterial (aluminum alloyed cast iron) under extra-low cycle rotating bending fatigue loading were studied using artificial neural networks (ANN) in this paper. The training data were used in the formation of training set of ANN. The ANN model exhibited excellent in results comparison with the experimental results. It was concluded that predicted fracture design parameters by the trained neural network model seem more reasonable compared to approximate methods. It is possible to claim that, ANN is fairly promising prediction technique if properly used. Training ANN model was introduced at first. And then the Training data for the development of the neural network model was obtained from the experiments. The input parameters, notch depth, the presetting deflection and tip radius of the notch, and the output parameters, the cycle times of fracture were used during the network training. The neural network architecture is designed. The ANN model was developed using back propagation architecture with three layers jump connections, where every layer was connected or linked to every previous layer. The number of hidden neurons was determined according to special formula. The performance of system is summarized at last. In order to facilitate the comparisons of predicted values, the error evaluation and mean relative error are obtained. The result show that the training model has good performance, and the experimental data and predicted data from ANN are in good coherence.

scholarly journals Bootstrapping a Neural Morphological Generator from Morphological Analyzer Output for Inuktitut

2019 ◽  
Vol 2 (1) ◽  
Author(s):  
Jeffrey Micher
Keyword(s):  
Neural Network ◽  
Training Data ◽  
Data Set ◽  
Set Size ◽  
The Neural Network ◽  
Surface Character ◽  
Finite State ◽  
Character Sequences ◽  
Finite State Transducer

We present a method for building a morphological generator from the output of an existing analyzer for Inuktitut, in the absence of a two-way finite state transducer which would normally provide this functionality. We make use of a sequence to sequence neural network which “translates” underlying Inuktitut morpheme sequences into surface character sequences. The neural network uses only the previous and the following morphemes as context. We report a morpheme accuracy of approximately 86%. We are able to increase this accuracy slightly by passing deep morphemes directly to output for unknown morphemes. We do not see significant improvement when increasing training data set size, and postulate possible causes for this.

Analysis of Fin-Tube Evaporator Performance With Limited Experimental Data Using Artificial Neural Networks

2000 ◽  
Author(s):  
Arturo Pacheco-Vega ◽  
Mihir Sen ◽  
Rodney L. McClain
Keyword(s):  
Neural Network ◽  
Heat Rate ◽  
Network Models ◽  
Activation Function ◽  
Operating Conditions ◽  
Training Data ◽  
Neural Network Models ◽  
The Neural Network ◽  
Artificial Neural ◽  
Fin Tube

Abstract In the current study we consider the problem of accuracy in heat rate estimations from artificial neural network models of heat exchangers used for refrigeration applications. The network configuration is of the feedforward type with a sigmoid activation function and a backpropagation algorithm. Limited experimental measurements from a manufacturer are used to show the capability of the neural network technique in modeling the heat transfer in these systems. Results from this exercise show that a well-trained network correlates the data with errors of the same order as the uncertainty of the measurements. It is also shown that the number and distribution of the training data are linked to the performance of the network when estimating the heat rates under different operating conditions, and that networks trained from few tests may give large errors. A methodology based on the cross-validation technique is presented to find regions where not enough data are available to construct a reliable neural network. The results from three tests show that the proposed methodology gives an upper bound of the estimated error in the heat rates.

scholarly journals Deep Convolutional Neural Network Phase Unwrapping for Fringe Projection 3D Imaging

Sensors ◽  
2020 ◽  
Vol 20 (13) ◽  
pp. 3691
Author(s):  
Jian Liang ◽  
Junchao Zhang ◽  
Jianbo Shao ◽  
Bofan Song ◽  
Baoli Yao ◽  
...  
Keyword(s):  
Neural Network ◽  
3D Imaging ◽  
Phase Unwrapping ◽  
Training Data ◽  
Fringe Projection ◽  
Second Step ◽  
Network Configuration ◽  
Simulation Data ◽  
The Neural Network ◽  
Key Features

Phase unwrapping is a very important step in fringe projection 3D imaging. In this paper, we propose a new neural network for accurate phase unwrapping to address the special needs in fringe projection 3D imaging. Instead of labeling the wrapped phase with integers directly, a two-step training process with the same network configuration is proposed. In the first step, the network (network I) is trained to label only four key features in the wrapped phase. In the second step, another network with same configuration (network II) is trained to label the wrapped phase segments. The advantages are that the dimension of the wrapped phase can be much larger from that of the training data, and the phase with serious Gaussian noise can be correctly unwrapped. We demonstrate the performance and key features of the neural network trained with the simulation data for the experimental data.

Prediction of stress in fillet portion of spur gears using artificial neural networks

2007 ◽  
Vol 22 (1) ◽  
pp. 41-51
Author(s):  
M.S. Shunmugam ◽  
N. Siva Prasad
Keyword(s):  
Neural Network ◽  
Gear Tooth ◽  
Spur Gear ◽  
Activation Function ◽  
Training Data ◽  
Spur Gears ◽  
Pressure Angle ◽  
The Neural Network ◽  
Artificial Neural ◽  
Taguchi’S Design Of Experiments

AbstractA fillet curve is provided at the root of the spur gear tooth, as stresses are high in this portion. The fillet curve may be a trochoid or an arc of suitable size as specified by designer. The fillet stress is influenced by the fillet geometry as well as the number of teeth, modules, and the pressure angle of the gear. Because the relationship is nonlinear and complex, an artificial neural network and a backpropagation algorithm are used in the present work to predict the fillet stresses. Training data are obtained from finite element simulations that are greatly reduced using Taguchi's design of experiments. Each simulation takes around 30 min. The 4-5-1 network and a sigmoid activation function are chosen. TRAINLM function is used for training the network with a learning rate parameter of 0.01 and a momentum constant of 0.8. The neural network is able to predict the fillet stresses in 0.03 s with reasonable accuracy for spur gears having 25–125 teeth, a 1–5 mm module, a 0.05–0.45 mm fillet radius, and a 15°–25° pressure angle.

scholarly journals Machine Learned Prediction of Reaction Template Applicability for Data-Driven Retrosynthetic Predictions of Energetic Materials

2020 ◽  
Author(s):  
Michael Fortunato ◽  
Connor W. Coley ◽  
Brian Barnes ◽  
Klavs F. Jensen
Keyword(s):  
Neural Network ◽  
Energetic Materials ◽  
Energetic Material ◽  
Training Data ◽  
Data Generation ◽  
Material Synthesis ◽  
Subgraph Matching ◽  
Planning Models ◽  
The Neural Network ◽  
Patent Literature

State of the art computer-aided synthesis planning models are naturally biased toward commonly reported chemical reactions, thus reducing the usefulness of those models for the unusual chemistry relevant to shock physics. To address this problem, a neural network was trained to recognize reaction template applicability for small organic molecules to supplement the rare reaction examples of relevance to energetic materials. The training data for the neural network was generated by brute force determination of template subgraph matching for product molecules from a database of reactions in U.S. patent literature. This data generation strategy successfully augmented the information about template applicability for rare reaction mechanisms in the reaction database. The increased ability to recognize rare reaction templates was demonstrated for reaction templates of interest for energetic material synthesis such as heterocycle ring formation.<div><br></div><div>The following article has been submitted to by the 21st Biennial APS Conference on Shock Compression of Condensed Matter. After it is published, it will be found at https://publishing.aip.org/resources/librarians/products/journals/.</div>

scholarly journals A Graph Neural Network for Predicting Energy and Stability of Known and Hypothetical Crystal Structures

2021 ◽  
Author(s):  
Shubham Pandey ◽  
Jiaxing Qu ◽  
Vladan Stevanovic ◽  
Peter St. John ◽  
Prashun Gorai
Keyword(s):  
Neural Network ◽  
Total Energy ◽  
Network Model ◽  
Neural Network Model ◽  
Density Functional ◽  
Failure Modes ◽  
Inorganic Materials ◽  
Training Data ◽  
The Neural Network ◽  
Unique Graph

The discovery of new inorganic materials in unexplored chemical spaces necessitates calculating total energy quickly and with sufficient accuracy. Machine learning models that provide such a capability for both ground-state (GS) and higher-energy structures would be instrumental in accelerating the screening for new materials over vast chemical spaces. Here, we develop a unique graph neural network model to accurately predict the total energy of both GS and higher-energy hypothetical structures. We use ~16,500 density functional theory calculated total energy from the NREL Materials Database and ~11,000 in-house generated hypothetical structures to train our model, thus making sure that the model is not biased towards either GS or higher-energy structures. We also demonstrate that our model satisfactorily ranks the structures in the correct order of their energies for a given composition. Furthermore, we present a thorough error analysis to explain several failure modes of the model, which highlights both prediction outliers and occasional inconsistencies in the training data. By peeling back layers of the neural network model, we are able to derive chemical trends by analyzing how the model represents learned structures and properties.

Sign in / Sign up

Export Citation Format

Share Document