scholarly journals High-Profile VRU Detection on Resource-Constrained Hardware Using YOLOv3/v4 on BDD100K

2020 ◽  
Vol 6 (12) ◽  
pp. 142
Author(s):  
Vicent Ortiz Ortiz Castelló ◽  
Ismael Salvador Salvador Igual ◽  
Omar del Tejo Catalá ◽  
Juan-Carlos Perez-Cortes
Keyword(s):  
Autonomous Vehicles ◽  
Data Augmentation ◽  
Activation Function ◽  
Road User ◽  
Activation Functions ◽  
Driver Assistance Systems ◽  
Trade Offs ◽  
Augmentation Techniques ◽  
Intrinsic Complexity ◽  
Detection Quality

Vulnerable Road User (VRU) detection is a major application of object detection with the aim of helping reduce accidents in advanced driver-assistance systems and enabling the development of autonomous vehicles. Due to intrinsic complexity present in computer vision and to limitations in processing capacity and bandwidth, this task has not been completely solved nowadays. For these reasons, the well established YOLOv3 net and the new YOLOv4 one are assessed by training them on a huge, recent on-road image dataset (BDD100K), both for VRU and full on-road classes, with a great improvement in terms of detection quality when compared to their MS-COCO-trained generic correspondent models from the authors but with negligible costs in forward pass time. Additionally, some models were retrained when replacing the original Leaky ReLU convolutional activation functions from original YOLO implementation with two cutting-edge activation functions: the self-regularized non-monotonic function (MISH) and its self-gated counterpart (SWISH), with significant improvements with respect to the original activation function detection performance. Additionally, some trials were carried out including recent data augmentation techniques (mosaic and cutmix) and some grid size configurations, with cumulative improvements over the previous results, comprising different performance-throughput trade-offs.

scholarly journals Deep Learning for Person Reidentification Using Support Vector Machines

2017 ◽  
Vol 2017 ◽  
pp. 1-12 ◽  
Author(s):  
Mengyu Xu ◽  
Zhenmin Tang ◽  
Yazhou Yao ◽  
Lingxiang Yao ◽  
Huafeng Liu ◽  
...  
Keyword(s):  
Data Augmentation ◽  
Metric Learning ◽  
Representation Learning ◽  
Activation Function ◽  
Feature Representation ◽  
Support Vector ◽  
Similarity Learning ◽  
Vector Machines ◽  
Augmentation Techniques ◽  
Suboptimal Solution

Due to the variations of viewpoint, pose, and illumination, a given individual may appear considerably different across different camera views. Tracking individuals across camera networks with no overlapping fields is still a challenging problem. Previous works mainly focus on feature representation and metric learning individually which tend to have a suboptimal solution. To address this issue, in this work, we propose a novel framework to do the feature representation learning and metric learning jointly. Different from previous works, we represent the pairs of pedestrian images as new resized input and use linear Support Vector Machine to replace softmax activation function for similarity learning. Particularly, dropout and data augmentation techniques are also employed in this model to prevent the network from overfitting. Extensive experiments on two publically available datasets VIPeR and CUHK01 demonstrate the effectiveness of our proposed approach.

scholarly journals Data Augmentation of Automotive LIDAR Point Clouds under Adverse Weather Situations

Sensors ◽  
2021 ◽  
Vol 21 (13) ◽  
pp. 4503
Author(s):  
Jose Roberto Vargas Rivero ◽  
Thiemo Gerbich ◽  
Boris Buschardt ◽  
Jia Chen
Keyword(s):  
Autonomous Vehicles ◽  
Data Augmentation ◽  
Synthetic Data ◽  
Real Data ◽  
Weather Conditions ◽  
Point Clouds ◽  
Training Data ◽  
Driver Assistance Systems ◽  
Solid Objects ◽  
Adverse Weather

In contrast to previous works on data augmentation using LIDAR (Light Detection and Ranging), which mostly consider point clouds under good weather conditions, this paper uses point clouds which are affected by spray. Spray water can be a cause of phantom braking and understanding how to handle the extra detections caused by it is an important step in the development of ADAS (Advanced Driver Assistance Systems)/AV (Autonomous Vehicles) functions. The extra detections caused by spray cannot be safely removed without considering cases in which real solid objects may be present in the same region in which the detections caused by spray take place. As collecting real examples would be extremely difficult, the use of synthetic data is proposed. Real scenes are reconstructed virtually with an added extra object in the spray region, in a way that the detections caused by this obstacle match the characteristics a real object in the same position would have regarding intensity, echo number and occlusion. The detections generated by the obstacle are then used to augment the real data, obtaining, after occlusion effects are added, a good approximation of the desired training data. This data is used to train a classifier achieving an average F-Score of 92. The performance of the classifier is analyzed in detail based on the characteristics of the synthetic object: size, position, reflection, duration. The proposed method can be easily expanded to different kinds of obstacles and classifier types.

Analysis of Non-Linear Activation Functions for Classification Tasks Using Convolutional Neural Networks

2019 ◽  
Vol 12 (3) ◽  
pp. 156-161 ◽  
Author(s):  
Aman Dureja ◽  
Payal Pahwa
Keyword(s):  
Neural Networks ◽  
Deep Neural Networks ◽  
Activation Function ◽  
Primary Objective ◽  
Experimental Comparison ◽  
Activation Functions ◽  
Practical Applications ◽  
Network Activation ◽  
Non Linear ◽  
Hidden Layer

Background: In making the deep neural network, activation functions play an important role. But the choice of activation functions also affects the network in term of optimization and to retrieve the better results. Several activation functions have been introduced in machine learning for many practical applications. But which activation function should use at hidden layer of deep neural networks was not identified. Objective: The primary objective of this analysis was to describe which activation function must be used at hidden layers for deep neural networks to solve complex non-linear problems. Methods: The configuration for this comparative model was used by using the datasets of 2 classes (Cat/Dog). The number of Convolutional layer used in this network was 3 and the pooling layer was also introduced after each layer of CNN layer. The total of the dataset was divided into the two parts. The first 8000 images were mainly used for training the network and the next 2000 images were used for testing the network. Results: The experimental comparison was done by analyzing the network by taking different activation functions on each layer of CNN network. The validation error and accuracy on Cat/Dog dataset were analyzed using activation functions (ReLU, Tanh, Selu, PRelu, Elu) at number of hidden layers. Overall the Relu gave best performance with the validation loss at 25th Epoch 0.3912 and validation accuracy at 25th Epoch 0.8320. Conclusion: It is found that a CNN model with ReLU hidden layers (3 hidden layers here) gives best results and improve overall performance better in term of accuracy and speed. These advantages of ReLU in CNN at number of hidden layers are helpful to effectively and fast retrieval of images from the databases.

scholarly journals Automatic Labeled Dialogue Generation for Nursing Record Systems

2020 ◽  
Vol 10 (3) ◽  
pp. 62
Author(s):  
Tittaya Mairittha ◽  
Nattaya Mairittha ◽  
Sozo Inoue
Keyword(s):  
Data Augmentation ◽  
Short Term Memory ◽  
Generative Models ◽  
Abstract Knowledge ◽  
Augmentation Techniques ◽  
Nursing Record ◽  
Long Short Term Memory ◽  
The Individual ◽  
High Level ◽  
Embedding Methods

The integration of digital voice assistants in nursing residences is becoming increasingly important to facilitate nursing productivity with documentation. A key idea behind this system is training natural language understanding (NLU) modules that enable the machine to classify the purpose of the user utterance (intent) and extract pieces of valuable information present in the utterance (entity). One of the main obstacles when creating robust NLU is the lack of sufficient labeled data, which generally relies on human labeling. This process is cost-intensive and time-consuming, particularly in the high-level nursing care domain, which requires abstract knowledge. In this paper, we propose an automatic dialogue labeling framework of NLU tasks, specifically for nursing record systems. First, we apply data augmentation techniques to create a collection of variant sample utterances. The individual evaluation result strongly shows a stratification rate, with regard to both fluency and accuracy in utterances. We also investigate the possibility of applying deep generative models for our augmented dataset. The preliminary character-based model based on long short-term memory (LSTM) obtains an accuracy of 90% and generates various reasonable texts with BLEU scores of 0.76. Secondly, we introduce an idea for intent and entity labeling by using feature embeddings and semantic similarity-based clustering. We also empirically evaluate different embedding methods for learning good representations that are most suitable to use with our data and clustering tasks. Experimental results show that fastText embeddings produce strong performances both for intent labeling and on entity labeling, which achieves an accuracy level of 0.79 and 0.78 f1-scores and 0.67 and 0.61 silhouette scores, respectively.

scholarly journals Hardware implementation of radial-basis neural networks with Gaussian activation functions on FPGA

2021 ◽  
Author(s):  
Volodymyr Shymkovych ◽  
Sergii Telenyk ◽  
Petro Kravets
Keyword(s):  
Neural Networks ◽  
Hardware Implementation ◽  
Gaussian Function ◽  
Activation Function ◽  
Rbf Neural Networks ◽  
Activation Functions ◽  
Rbf Network ◽  
Combination Scheme ◽  
Radial Basis ◽  
Hidden Layer

AbstractThis article introduces a method for realizing the Gaussian activation function of radial-basis (RBF) neural networks with their hardware implementation on field-programmable gaits area (FPGAs). The results of modeling of the Gaussian function on FPGA chips of different families have been presented. RBF neural networks of various topologies have been synthesized and investigated. The hardware component implemented by this algorithm is an RBF neural network with four neurons of the latent layer and one neuron with a sigmoid activation function on an FPGA using 16-bit numbers with a fixed point, which took 1193 logic matrix gate (LUTs—LookUpTable). Each hidden layer neuron of the RBF network is designed on an FPGA as a separate computing unit. The speed as a total delay of the combination scheme of the block RBF network was 101.579 ns. The implementation of the Gaussian activation functions of the hidden layer of the RBF network occupies 106 LUTs, and the speed of the Gaussian activation functions is 29.33 ns. The absolute error is ± 0.005. The Spartan 3 family of chips for modeling has been used to get these results. Modeling on chips of other series has been also introduced in the article. RBF neural networks of various topologies have been synthesized and investigated. Hardware implementation of RBF neural networks with such speed allows them to be used in real-time control systems for high-speed objects.

scholarly journals Generative Adversarial Networks to Improve the Robustness of Visual Defect Segmentation by Semantic Networks in Manufacturing Components

2021 ◽  
Vol 11 (14) ◽  
pp. 6368
Author(s):  
Fátima A. Saiz ◽  
Garazi Alfaro ◽  
Iñigo Barandiaran ◽  
Manuel Graña
Keyword(s):  
Ad Hoc ◽  
Data Augmentation ◽  
Semantic Network ◽  
Semantic Networks ◽  
Stereo Image ◽  
Generative Adversarial Networks ◽  
Specific Class ◽  
Adversarial Networks ◽  
Augmentation Techniques ◽  
Image Acquisition System

This paper describes the application of Semantic Networks for the detection of defects in images of metallic manufactured components in a situation where the number of available samples of defects is small, which is rather common in real practical environments. In order to overcome this shortage of data, the common approach is to use conventional data augmentation techniques. We resort to Generative Adversarial Networks (GANs) that have shown the capability to generate highly convincing samples of a specific class as a result of a game between a discriminator and a generator module. Here, we apply the GANs to generate samples of images of metallic manufactured components with specific defects, in order to improve training of Semantic Networks (specifically DeepLabV3+ and Pyramid Attention Network (PAN) networks) carrying out the defect detection and segmentation. Our process carries out the generation of defect images using the StyleGAN2 with the DiffAugment method, followed by a conventional data augmentation over the entire enriched dataset, achieving a large balanced dataset that allows robust training of the Semantic Network. We demonstrate the approach on a private dataset generated for an industrial client, where images are captured by an ad-hoc photometric-stereo image acquisition system, and a public dataset, the Northeastern University surface defect database (NEU). The proposed approach achieves an improvement of 7% and 6% in an intersection over union (IoU) measure of detection performance on each dataset over the conventional data augmentation.

scholarly journals Data Augmentation Techniques on Arabic Data for Named Entity Recognition

2021 ◽  
Vol 189 ◽  
pp. 292-299
Author(s):  
Caroline Sabty ◽  
Islam Omar ◽  
Fady Wasfalla ◽  
Mohamed Islam ◽  
Slim Abdennadher
Keyword(s):  
Data Augmentation ◽  
Named Entity Recognition ◽  
Entity Recognition ◽  
Named Entity ◽  
Augmentation Techniques

scholarly journals Trigonometric Inference Providing Learning in Deep Neural Networks

2021 ◽  
Vol 11 (15) ◽  
pp. 6704
Author(s):  
Jingyong Cai ◽  
Masashi Takemoto ◽  
Yuming Qiu ◽  
Hironori Nakajo
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Deep Neural Networks ◽  
Activation Function ◽  
Trigonometric Approximation ◽  
Model Parameters ◽  
Training Algorithms ◽  
Activation Functions ◽  
Classical Training ◽  
Sum Formula

Despite being heavily used in the training of deep neural networks (DNNs), multipliers are resource-intensive and insufficient in many different scenarios. Previous discoveries have revealed the superiority when activation functions, such as the sigmoid, are calculated by shift-and-add operations, although they fail to remove multiplications in training altogether. In this paper, we propose an innovative approach that can convert all multiplications in the forward and backward inferences of DNNs into shift-and-add operations. Because the model parameters and backpropagated errors of a large DNN model are typically clustered around zero, these values can be approximated by their sine values. Multiplications between the weights and error signals are transferred to multiplications of their sine values, which are replaceable with simpler operations with the help of the product to sum formula. In addition, a rectified sine activation function is utilized for further converting layer inputs into sine values. In this way, the original multiplication-intensive operations can be computed through simple add-and-shift operations. This trigonometric approximation method provides an efficient training and inference alternative for devices with insufficient hardware multipliers. Experimental results demonstrate that this method is able to obtain a performance close to that of classical training algorithms. The approach we propose sheds new light on future hardware customization research for machine learning.

Sign in / Sign up

Export Citation Format

Share Document