scholarly journals Convolutional Neural Network Using Kalman Filter for Human Detection and Tracking on RGB-D Video

2018 ◽  
Vol 12 (2) ◽  
pp. 105 ◽  
Author(s):  
Jovin Angelico ◽  
Ken Ratri Retno Wardani
Keyword(s):  
Neural Network ◽  
Kalman Filter ◽  
Convolutional Neural Network ◽  
Human Detection ◽  
Detection Accuracy ◽  
Tracking Accuracy ◽  
Lighting Condition ◽  
Detection And Tracking ◽  
Computation Efficiency ◽  
The Impact

The computer ability to detect human being by computer vision is still being improved both in accuracy or computation time. In low-lighting condition, the detection accuracy is usually low. This research uses additional information, besides RGB channels, namely a depth map that shows objects’ distance relative to the camera. This research integrates Cascade Classifier (CC) to localize the potential object, the Convolutional Neural Network (CNN) technique to identify the human and nonhuman image, and the Kalman filter technique to track human movement. For training and testing purposes, there are two kinds of RGB-D datasets used with different points of view and lighting conditions. Both datasets have been selected to remove images which contain a lot of noises and occlusions so that during the training process it will be more directed. Using these integrated techniques, detection and tracking accuracy reach 77.7%. The impact of using Kalman filter increases computation efficiency by 41%.

scholarly journals Detection of Atrial Fibrillation Using 1D Convolutional Neural Network

Sensors ◽  
2020 ◽  
Vol 20 (7) ◽  
pp. 2136 ◽  
Author(s):  
Chaur-Heh Hsieh ◽  
Yan-Shuo Li ◽  
Bor-Jiunn Hwang ◽  
Ching-Hua Hsiao
Keyword(s):  
Neural Network ◽  
Atrial Fibrillation ◽  
Convolutional Neural Network ◽  
Detection System ◽  
Detection Methods ◽  
Detection Accuracy ◽  
Implementation Cost ◽  
Electrocardiogram Ecg ◽  
The Impact ◽  
Heavy Training

The automatic detection of atrial fibrillation (AF) is crucial for its association with the risk of embolic stroke. Most of the existing AF detection methods usually convert 1D time-series electrocardiogram (ECG) signal into 2D spectrogram to train a complex AF detection system, which results in heavy training computation and high implementation cost. This paper proposes an AF detection method based on an end-to-end 1D convolutional neural network (CNN) architecture to raise the detection accuracy and reduce network complexity. By investigating the impact of major components of a convolutional block on detection accuracy and using grid search to obtain optimal hyperparameters of the CNN, we develop a simple, yet effective 1D CNN. Since the dataset provided by PhysioNet Challenge 2017 contains ECG recordings with different lengths, we also propose a length normalization algorithm to generate equal-length records to meet the requirement of CNN. Experimental results and analysis indicate that our method of 1D CNN achieves an average F1 score of 78.2%, which has better detection accuracy with lower network complexity, as compared with the existing deep learning-based methods.

Automatic Monitoring of Chicken Movement and Drinking Time Using Convolutional Neural Networks

2020 ◽  
Vol 63 (6) ◽  
pp. 2029-2038
Author(s):  
Chen-Yi Lin ◽  
Kuang-Wen Hsieh ◽  
Yao-Chuan Tsai ◽  
Yan-Fu Kuo
Keyword(s):  
Neural Network ◽  
Heat Stress ◽  
Embedded System ◽  
Convolutional Neural Network ◽  
Time Lapse ◽  
Detection Accuracy ◽  
List Type ◽  
Tracking Accuracy ◽  
Temperature Humidity Index ◽  
Humidity Index

HighlightsA customized embedded system was built to acquire images of a chicken coop.Faster R-CNN was used to localize the chickens in the images.The accuracies in chicken detection and tracking were 98.16% and 98.94%, respectively.Movement and drinking time of chickens were quantified.Abstract. Poultry and eggs are major sources of dietary protein worldwide. Because Taiwan is located in tropical and subtropical regions, heat stress in chickens is one of the most challenging concerns of the poultry industry in Taiwan. Typical heat stress symptoms in chickens are reduced movement and increased drinking time. The level of heat stress is conventionally evaluated using the temperature-humidity index (THI) or through manual observation. However, THI is indirect, and manual observation is subjective and time-consuming. This study proposes to directly monitor the movement and drinking time of chickens using time-lapse images and deep learning algorithms. In this study, an experimental coop was constructed to house ten chickens. An embedded system was then designed to acquire images of the chickens at a rate of 1 frame s-1 and to measure the temperature and humidity of the coop. A faster region-based convolutional neural network was then trained on a personal computer to detect and localize the chickens in the images. The movement and drinking time of the chickens under various THI values were then analyzed. The proposed method provided 98.16% chicken detection accuracy and 98.94% chicken tracking accuracy. Keywords: Chicken activities, Embedded system, Faster region-based convolutional neural network, Faster R-CNN, Heat stress, Temperature-humidity index (THI).

Deep convolutional neural networks for human movement detection using wireless signals

2021 ◽  
pp. 1-10
Author(s):  
Chien-Cheng Leea ◽  
Zhongjian Gao ◽  
Xiu-Chi Huanga
Keyword(s):  
Neural Network ◽  
Feature Extraction ◽  
Convolutional Neural Network ◽  
Detection System ◽  
Deep Convolutional Neural Network ◽  
Human Detection ◽  
Two Dimensional ◽  
Dimensional Matrix ◽  
State Classification ◽  
Propagation Paths

This paper proposes a Wi-Fi-based indoor human detection system using a deep convolutional neural network. The system detects different human states in various situations, including different environments and propagation paths. The main improvements proposed by the system is that there is no cameras overhead and no sensors are mounted. This system captures useful amplitude information from the channel state information and converts this information into an image-like two-dimensional matrix. Next, the two-dimensional matrix is used as an input to a deep convolutional neural network (CNN) to distinguish human states. In this work, a deep residual network (ResNet) architecture is used to perform human state classification with hierarchical topological feature extraction. Several combinations of datasets for different environments and propagation paths are used in this study. ResNet’s powerful inference simplifies feature extraction and improves the accuracy of human state classification. The experimental results show that the fine-tuned ResNet-18 model has good performance in indoor human detection, including people not present, people still, and people moving. Compared with traditional machine learning using handcrafted features, this method is simple and effective.

scholarly journals Activities of daily living with bionic arm improved by combination training and latching filter in prosthesis control comparison

2021 ◽  
Vol 18 (1) ◽  
Author(s):  
Michael D. Paskett ◽  
Mark R. Brinton ◽  
Taylor C. Hansen ◽  
Jacob A. George ◽  
Tyler S. Davis ◽  
...  
Keyword(s):  
Neural Network ◽  
Kalman Filter ◽  
Activities Of Daily Living ◽  
Convolutional Neural Network ◽  
Daily Living ◽  
Complex Data ◽  
Multi Layer Perceptron ◽  
Experimental Conditions ◽  
Modified Kalman Filter ◽  
Prosthesis Control

Abstract Background Advanced prostheses can restore function and improve quality of life for individuals with amputations. Unfortunately, most commercial control strategies do not fully utilize the rich control information from residual nerves and musculature. Continuous decoders can provide more intuitive prosthesis control using multi-channel neural or electromyographic recordings. Three components influence continuous decoder performance: the data used to train the algorithm, the algorithm, and smoothing filters on the algorithm’s output. Individual groups often focus on a single decoder, so very few studies compare different decoders using otherwise similar experimental conditions. Methods We completed a two-phase, head-to-head comparison of 12 continuous decoders using activities of daily living. In phase one, we compared two training types and a smoothing filter with three algorithms (modified Kalman filter, multi-layer perceptron, and convolutional neural network) in a clothespin relocation task. We compared training types that included only individual digit and wrist movements vs. combination movements (e.g., simultaneous grasp and wrist flexion). We also compared raw vs. nonlinearly smoothed algorithm outputs. In phase two, we compared the three algorithms in fragile egg, zipping, pouring, and folding tasks using the combination training and smoothing found beneficial in phase one. In both phases, we collected objective, performance-based (e.g., success rate), and subjective, user-focused (e.g., preference) measures. Results Phase one showed that combination training improved prosthesis control accuracy and speed, and that the nonlinear smoothing improved accuracy but generally reduced speed. Phase one importantly showed simultaneous movements were used in the task, and that the modified Kalman filter and multi-layer perceptron predicted more simultaneous movements than the convolutional neural network. In phase two, user-focused metrics favored the convolutional neural network and modified Kalman filter, whereas performance-based metrics were generally similar among all algorithms. Conclusions These results confirm that state-of-the-art algorithms, whether linear or nonlinear in nature, functionally benefit from training on more complex data and from output smoothing. These studies will be used to select a decoder for a long-term take-home trial with implanted neuromyoelectric devices. Overall, clinical considerations may favor the mKF as it is similar in performance, faster to train, and computationally less expensive than neural networks.

scholarly journals Applying of machine learning in the construction of a voice-controlled interface on the example of a music player

2019 ◽  
Vol 13 ◽  
pp. 302-309
Author(s):  
Jakub Basiakowski
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Convolutional Neural Network ◽  
Feedforward Neural Network ◽  
Hidden Layer ◽  
Music Player ◽  
The Impact

The following paper presents the results of research on the impact of machine learning in the construction of a voice-controlled interface. Two different models were used for the analysys: a feedforward neural network containing one hidden layer and a more complicated convolutional neural network. What is more, a comparison of the applied models was presented. This comparison was performed in terms of quality and the course of training.

scholarly journals RICNN: A ResNet&Inception convolutional neural network for intrusion detection of abnormal traffic

2021 ◽  
pp. 55-55
Author(s):  
Benhui Xia ◽  
Dezhi Han ◽  
Ximing Yin ◽  
Gao Na
Keyword(s):  
Neural Network ◽  
Intrusion Detection ◽  
Convolutional Neural Network ◽  
Detection Rate ◽  
Recognition Rate ◽  
Identification Accuracy ◽  
Detection Accuracy ◽  
Traffic Classification ◽  
High Detection Rate ◽  
Network Intrusion

To secure cloud computing and outsourced data while meeting the requirements of automation, many intrusion detection schemes based on deep learn ing are proposed. Though the detection rate of many network intrusion detection solutions can be quite high nowadays, their identification accuracy on imbalanced abnormal network traffic still remains low. Therefore, this paper proposes a ResNet &Inception-based convolutional neural network (RICNN) model to abnormal traffic classification. RICNN can learn more traffic features through the Inception unit, and the degradation problem of the network is eliminated through the direct map ping unit of ResNet, thus the improvement of the model?s generalization ability can be achievable. In addition, to simplify the network, an improved version of RICNN, which makes it possible to reduce the number of parameters that need to be learnt without degrading identification accuracy, is also proposed in this paper. The experimental results on the dataset CICIDS2017 show that RICNN not only achieves an overall accuracy of 99.386% but also has a high detection rate across different categories, especially for small samples. The comparison experiments show that the recognition rate of RICNN outperforms a variety of CNN models and RNN models, and the best detection accuracy can be achieved.

scholarly journals A Single Target Grasp Detection Network Based on Convolutional Neural Network

2021 ◽  
Vol 2021 ◽  
pp. 1-12
Author(s):  
Longzhi Zhang ◽  
Dongmei Wu
Keyword(s):  
Neural Network ◽  
Convolutional Neural Network ◽  
Transfer Learning ◽  
High Accuracy ◽  
Experimental Results ◽  
Detection Accuracy ◽  
Single Object ◽  
Single Target ◽  
High Detection ◽  
Good Detection

Grasp detection based on convolutional neural network has gained some achievements. However, overfitting of multilayer convolutional neural network still exists and leads to poor detection precision. To acquire high detection accuracy, a single target grasp detection network that generalizes the fitting of angle and position, based on the convolution neural network, is put forward here. The proposed network regards the image as input and grasping parameters including angle and position as output, with the detection manner of end-to-end. Particularly, preprocessing dataset is to achieve the full coverage to input of model and transfer learning is to avoid overfitting of network. Importantly, a series of experimental results indicate that, for single object grasping, our network has good detection results and high accuracy, which proves that the proposed network has strong generalization in direction and category.

Sign in / Sign up

Export Citation Format

Share Document