• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.6
• /
• pp.842-849
• /
• 2022

With the development of deep learning technology, there are many cases of using DNNs in embedded systems such as unmanned vehicles, drones, and robotics. Typically, in the case of an autonomous driving system, it is crucial to run several DNNs which have high accuracy results and large computation amount at the same time. However, running multiple DNNs simultaneously in an embedded system with relatively low performance increases the time required for the inference. This phenomenon may cause a problem of performing an abnormal function because the operation according to the inference result is not performed in time. To solve this problem, the solution proposed in this paper first reduces the computation by applying the Tucker decomposition to DNN models with big computation amount, and then, make DNN models run in parallel as much as possible in the unit of hidden layer inside the GPU. The experimental result shows that the DNN inference time decreases by up to 75.6% compared to the case before applying the proposed technique.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.67-69
• /
• 2023

유연한 이동성, 쉬운 배치, 저렴한 비용 등의 장점을 가진 Unmanned Aerial Vehicle(UAV)를 이용해 Deep Neural Network(DNN) 서비스를 제공하는 기술이 연구되고 있다. 하지만 UAV 는 메모리와 컴퓨팅 능력, 배터리가 제한되어 있어 DNN 서비스의 요구사항을 만족시키기 위해서는 다수의 UAV간의 협업이 필요하다. 본 논문에서는 다수의 UAV 협업 환경에서 DNN 서비스의 처리 지연시간을 줄이기 위해 UAV 들의 작업량을 고려한 서비스 분산 처리 기법을 제안하고, 시뮬레이션을 통해 DNN 서비스 처리 지연 시간을 분석한다.

• Phonetics and Speech Sciences
• /
• v.9 no.2
• /
• pp.95-101
• /
• 2017

This paper proposes a new method to train Deep Neural Network (DNN)-based acoustic models for speech recognition of native and foreign speakers. The proposed method consists of determining multi-set state clusters with various acoustic properties, training a DNN-based acoustic model, and recognizing speech based on the model. In the proposed method, hidden nodes of DNN are shared, but output nodes are separated to accommodate different acoustic properties for native and foreign speech. In an English speech recognition task for speakers of Korean and English respectively, the proposed method is shown to slightly improve recognition accuracy compared to the conventional multi-condition training method.

• Journal of KIISE
• /
• v.43 no.11
• /
• pp.1281-1297
• /
• 2016

In recent years, personal videos have seen a tremendous growth due to the substantial increase in the use of smart devices and networking services in which users create and share video content easily without many restrictions. However, taking both into account would significantly improve event detection performance because videos generally have multiple modalities and the frame data in video varies at different time points. This paper proposes an event detection method. In this method, high-level features are first extracted from multiple modalities in the videos, and the features are rearranged according to time sequence. Then the association of the modalities is learned by means of DNN to produce a personal video event detector. In our proposed method, audio and image data are first synchronized and then extracted. Then, the result is input into GoogLeNet as well as Multi-Layer Perceptron (MLP) to extract high-level features. The results are then re-arranged in time sequence, and every video is processed to extract one feature each for training by means of DNN.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.6
• /
• pp.714-721
• /
• 2020

DNN is expanding its use in embedded systems such as robots and autonomous vehicles. For high recognition accuracy, computational complexity is greatly increased, and multiple DNNs are running aperiodically. Therefore, the ability processing multiple DNNs in embedded environments is a crucial issue. Accordingly, multicore based platforms are being released. However, most DNN models are operated in a batch process, and when multiple DNNs are operated in multicore together, the execution time deviation between each DNN may be large and the end-to-end execution time of the whole DNNs could be long depending on how they are allocated to the cores. In this paper, we solve these problems by providing a framework that decompose each DNN into individual layers and then distribute to multicores through a global queue. As a result of the experiment, the total DNN execution time was reduced by 31%, and when operating multiple identical DNNs, the deviation in execution time was reduced by up to 95.1%.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.3
• /
• pp.368-376
• /
• 2021

With the recent development of deep learning technology, autonomous things technology is attracting attention, and DNNs are widely used in embedded systems such as drones and autonomous vehicles. Embedded systems that can perform large-scale operations and process multiple DNNs for high recognition accuracy without relying on the cloud are being released. DNNs with various levels of priority exist within these systems. DNNs related to the safety-critical applications of autonomous vehicles have the highest priority, and they must be handled first. In this paper, we propose a priority-based scheduling framework for DNNs when multiple DNNs are executed simultaneously. Even if a low-priority DNN is being executed first, a high-priority DNN can preempt it, guaranteeing the fast response characteristics of safety-critical applications of autonomous vehicles. As a result of checking through extensive experiments, the performance improved by up to 76.6% in the actual commercial board.

• Journal of Internet of Things and Convergence
• /
• v.9 no.2
• /
• pp.71-76
• /
• 2023

In this paper, we propose a smoothed received signal strength indicator (RSSI)-based distance estimation using deep neural network (DNN) for accurate distance estimation in an environment where a single receiver is used. The proposed scheme performs a data preprocessing consisting of data splitting, missing value imputation, and smoothing steps to improve distance estimation accuracy, thereby deriving the smoothed RSSI values. The derived smoothed RSSI values are used as input data of the Multi-Input Single-Output (MISO) DNN model, and are finally returned as an estimated distance in the output layer through input layer and hidden layer. To verify the superiority of the proposed scheme, we compared the performance of the proposed scheme with that of the linear regression-based distance estimation scheme. As a result, the proposed scheme showed 29.09% higher distance estimation accuracy than the linear regression-based distance estimation scheme.

• Journal of Broadcast Engineering
• /
• v.23 no.3
• /
• pp.361-368
• /
• 2018

As autonomous driving research is rapidly developing, pedestrian detection study is also successfully investigated. However, most of the study utilizes color image datasets and those are relatively easy to detect the pedestrian. In case of color images, the scene should be exposed by enough light in order to capture the pedestrian and it is not easy for the conventional methods to detect the pedestrian if it is the other case. Therefore, in this paper, we propose deep neural network (DNN)-based multi-spectrum pedestrian detection method using color and thermal images. Based on single-shot multibox detector (SSD), we propose fusion network structures which simultaneously employ color and thermal images. In the experiment, we used KAIST dataset. We showed that proposed SSD-H (SSD-Halfway fusion) technique shows 18.18% lower miss rate compared to the KAIST pedestrian detection baseline. In addition, the proposed method shows at least 2.1% lower miss rate compared to the conventional halfway fusion method.

• ETRI Journal
• /
• v.39 no.3
• /
• pp.398-405
• /
• 2017

In this paper, an artificial stereo extension method that creates stereophonic sound from a mono sound source is proposed. The proposed method first trains deep neural networks (DNNs) that model the nonlinear relationship between the dominant and residual signals of the stereo channel. In the training stage, the band-wise log spectral magnitude and unwrapped phase of both the dominant and residual signals are utilized to model the nonlinearities of each sub-band through deep architecture. From that point, stereo extension is conducted by estimating the residual signal that corresponds to the input mono channel signal with the trained DNN model in a sub-band domain. The performance of the proposed method was evaluated using a log spectral distortion (LSD) measure and multiple stimuli with a hidden reference and anchor (MUSHRA) test. The results showed that the proposed method provided a lower LSD and higher MUSHRA score than conventional methods that use hidden Markov models and DNN with full-band processing.

• ETRI Journal
• /
• v.42 no.5
• /
• pp.658-668
• /
• 2020

In edge computing, most procedures, including data collection, data processing, and service provision, are handled at edge nodes and not in the central cloud. This decreases the processing burden on the central cloud, enabling fast responses to end-device service requests in addition to reducing bandwidth consumption. However, edge nodes have restricted computing, storage, and energy resources to support computation-intensive tasks such as processing deep neural network (DNN) inference. In this study, we analyze the effect of models with single and multiple local exits on DNN inference in an edge-computing environment. Our test results show that a single-exit model performs better with respect to the number of local exited samples, inference accuracy, and inference latency than a multi-exit model at all exit points. These results signify that higher accuracy can be achieved with less computation when a single-exit model is adopted. In edge computing infrastructure, it is therefore more efficient to adopt a DNN model with only one or a few exit points to provide a fast and reliable inference service.