• IEMEK Journal of Embedded Systems and Applications
• /
• v.19 no.2
• /
• pp.73-81
• /
• 2024

Recently, mobile manipulators have been utilized in agriculture industry for weed removal and harvest automation. This paper proposes a domain adaptive fruit detection method for harvest automation, by utilizing OWL-ViT model which is an open-vocabulary object detection model. The vision-language model can detect objects based on text prompt, and therefore, it can be extended to detect objects of undefined categories. In the development of deep learning models for real-world problems, constructing a large-scale labeled dataset is a time-consuming task and heavily relies on human effort. To reduce the labor-intensive workload, we utilized a large-scale public dataset as a source domain data and employed a domain adaptation method. Adversarial learning was conducted between a domain discriminator and feature extractor to reduce the gap between the distribution of feature vectors from the source domain and our target domain data. We collected a target domain dataset in a real-like environment and conducted experiments to demonstrate the effectiveness of the proposed method. In experiments, the domain adaptation method improved the AP50 metric from 38.88% to 78.59% for detecting objects within the range of 2m, and we achieved 81.7% of manipulation success rate.

• The Journal of Korea Robotics Society
• /
• v.13 no.1
• /
• pp.1-7
• /
• 2018

The image-to-image translation is one of the deep learning applications using image data. In this paper, we aim at improving the performance of object transfiguration which transforms a specific object in an image into another specific object. For object transfiguration, it is required to transform only the target object and maintain background images. In the existing results, however, it is observed that other parts in the image are also transformed. In this paper, we have focused on the structure of artificial neural networks that are frequently used in the existing methods and have improved the performance by adding constraints to the exiting structure. We also propose the advanced structure that combines the existing structures to maintain their advantages and complement their drawbacks. The effectiveness of the proposed methods are shown in experimental results.

• Journal of applied mathematics & informatics
• /
• v.39 no.1_2
• /
• pp.239-249
• /
• 2021

This study attempts to predict the price of gold futures, a real financial product, using ARIMA and LSTM. The wavelet analysis was applied to the data to predict the price of gold futures through LSTM and ARIMA. As results, it is confirmed that the prediction performance of the existing model of predict was improved. the case of predict of price of gold futures, we confirmed that the use of a deep learning model that is not affected by the non-stationary series data is suitable and the possibility of improving the accuracy of prediction through wavelet analysis.

• Journal of the Korean Society of Industry Convergence
• /
• v.25 no.4_2
• /
• pp.573-586
• /
• 2022

Corrosion can cause dangerous and expensive damage and failures of ship hulls and equipment. Therefore, it is necessary to maintain the vessel by periodic corrosion inspections. During visual inspection, many corrosion locations are inaccessible for many reasons, especially safety's point of view. Including subjective decisions of inspectors is one of the issues of visual inspection. Automation of visual inspection is tried by many pieces of research. In this study, we propose image preprocessing methods by image patch segmentation and thresholding. YOLOv5 was used as an object detection model after the image preprocessing. Finally, it was evaluated that corrosion detection performance using the proposed method was improved in terms of mean average precision.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.553-554
• /
• 2021

This paper proposes a convergence model based on LSTM and CNN deep learning techniques, and obtains good results by applying it to multi-category news datasets. According to the experiment, the deep learning-based fusion model significantly improved the precision and accuracy of text sentiment classification.

• Journal of Information Processing Systems
• /
• v.18 no.6
• /
• pp.803-812
• /
• 2022

With the increasing number of mobile device users worldwide, utilizing mobile edge computing (MEC) devices close to users for content caching can reduce transmission latency than receiving content from a server or cloud. However, because MEC has limited storage capacity, it is necessary to determine the content types and sizes to be cached. In this study, we investigate a caching strategy that increases the hit ratio from small base stations (SBSs) for mobile users in a heterogeneous network consisting of one macro base station (MBS) and multiple SBSs. If there are several SBSs that users can access, the hit ratio can be improved by reducing duplicate content and increasing the diversity of content in SBSs. We propose a Deep Q-Network (DQN)-based caching strategy that considers time-varying content popularity and content redundancy in multiple SBSs. Content is stored in the SBS in a divided form using maximum distance separable (MDS) codes to enhance the diversity of the content. Experiments in various environments show that the proposed caching strategy outperforms the other methods in terms of hit ratio.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.493-494
• /
• 2022

The development of livestock faces intensive farming results in a rising need for recognition of individual animals such as cows and pigs is related to high traceability. In this paper, we present a non-invasive biometrics systematic approach based on the deep-learning classification model to pig face identification. Firstly, in our systematic method, we build a ROS data collection system block to collect 10 pig face data images. Secondly, we proposed a preprocessing block in that we utilize the SSIM method to filter some images of collected images that have high similarity. Thirdly, we employ the improved image classification model of CNN (ViT), which uses the finetuning and pretraining technique to recognize the individual pig face. Finally, our proposed method achieves the accuracy about 98.66%.

• Journal of Advanced Navigation Technology
• /
• v.21 no.4
• /
• pp.365-370
• /
• 2017

This study analyze correlation between weekdays data and special days data of different power demand patterns, and builds a separate data set, and suggests ways to reduce power demand prediction error by using deep learning network suitable for each data set. In addition, we propose a method to improve the prediction rate by adding the environmental elements and the separating element to the meteorological element, which is a basic power demand prediction elements. The entire data predicted power demand using LSTM which is suitable for learning time series data, and the special day data predicted power demand using DNN. The experiment result show that the prediction rate is improved by adding prediction elements other than meteorological elements. The average RMSE of the entire dataset was 0.2597 for LSTM and 0.5474 for DNN, indicating that the LSTM showed a good prediction rate. The average RMSE of the special day data set was 0.2201 for DNN, indicating that the DNN had better prediction than LSTM. The MAPE of the LSTM of the whole data set was 2.74% and the MAPE of the special day was 3.07 %.

• Journal of Korea Multimedia Society
• /
• v.25 no.10
• /
• pp.1375-1385
• /
• 2022

In this paper, we propose a semi-supervised domain adaptation solution to deal with practical face recognition (FR) scenarios where a single face image for each target identity (to be recognized) is only available in the training phase. Main goal of the proposed method is to reduce the discrepancy between the target and the source domain face images, which ultimately improves FR performances. The proposed method is based on the Domain Adatation network (DAN) using an MMD loss function to reduce the discrepancy between domains. In order to train more effectively, we develop a novel loss function learning strategy in which MMD loss and cross-entropy loss functions are adopted by using different weights according to the progress of each epoch during the learning. The proposed weight adoptation focuses on the training of the source domain in the initial learning phase to learn facial feature information such as eyes, nose, and mouth. After the initial learning is completed, the resulting feature information is used to training a deep network using the target domain images. To evaluate the effectiveness of the proposed method, FR performances were evaluated with pretrained model trained only with CASIA-webface (source images) and fine-tuned model trained only with FERET's gallery (target images) under the same FR scenarios. The experimental results showed that the proposed semi-supervised domain adaptation can be improved by 24.78% compared to the pre-trained model and 28.42% compared to the fine-tuned model. In addition, the proposed method outperformed other state-of-the-arts domain adaptation approaches by 9.41%.

• Journal of Multimedia Information System
• /
• v.9 no.2
• /
• pp.93-102
• /
• 2022

Dependency parsing is a decision problem of the syntactic relation between words in a sentence. Recently, deep learning models are used for dependency parsing based on the word representations in a continuous vector space. However, it causes a mislabeled tagging problem for the proper nouns that rarely appear in the training corpus because it is difficult to express out-of-vocabulary (OOV) words in a continuous vector space. To solve the OOV problem in dependency parsing, we explored the proper noun embedding method according to the embedding unit. Before representing words in a continuous vector space, we replace the proper nouns with a special token and train them for the contextual features by using the multi-layer bidirectional LSTM. Two models of the syllable-based and morpheme-based unit are proposed for proper noun embedding and the performance of the dependency parsing is more improved in the ensemble model than each syllable and morpheme embedding model. The experimental results showed that our ensemble model improved 1.69%p in UAS and 2.17%p in LAS than the same arc-eager approach-based Malt parser.