http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Cable Instance Segmentation with Synthetic Data Generation
Assefa Seyoum Wahd,Donghyung Kim,Seung-Ik Lee 제어로봇시스템학회 2022 제어로봇시스템학회 국제학술대회 논문집 Vol.2022 No.11
We propose a bottom-up approach for the instance segmentation of cables (commonly referred in the literature as deformable linear objects). While the state of the art instance segmentation techniques propose a bounding box and perform foreground segmentation within each proposed bounding box, we adopt a bottom-up approach as cables can span a considerable part of the image or even the entire image, and therefore, cannot be well localized in a bounding box. In this paper, we show that several operations in the top-down instance segmentation approaches are only applicable for certain classes (i.e., compact objects) such as cars but they are a poor approximation for objects with highly overlapping bounding boxes such as cables. In particular, the non-maximum suppression and RoIPool/RoIAlign operations limit the generalizability of proposal-based instance segmentation methods to such datasets. Furthermore, we introduce a synthetic data generation technique that can also be applied to other popular public datasets such as COCO, Pascal VOC, and Cityscapes.
딥러닝을 사용하는 IoT빅데이터 인프라에 필요한 DNA 기술을 위한 분산 엣지 컴퓨팅기술 리뷰
( Temesgen Seyoum Alemayehu ),조위덕 ( We-duke Cho ) 한국정보처리학회 2020 정보처리학회논문지. 컴퓨터 및 통신시스템 Vol.9 No.12
Nowadays, Data-Network-AI (DNA)-based intelligent services and applications have become a reality to provide a new dimension of services that improve the quality of life and productivity of businesses. Artificial intelligence (AI) can enhance the value of IoT data (data collected by IoT devices). The internet of things (IoT) promotes the learning and intelligence capability of AI. To extract insights from massive volume IoT data in real-time using deep learning, processing capability needs to happen in the IoT end devices where data is generated. However, deep learning requires a significant number of computational resources that may not be available at the IoT end devices. Such problems have been addressed by transporting bulks of data from the IoT end devices to the cloud datacenters for processing. But transferring IoT big data to the cloud incurs prohibitively high transmission delay and privacy issues which are a major concern. Edge computing, where distributed computing nodes are placed close to the IoT end devices, is a viable solution to meet the high computation and low-latency requirements and to preserve the privacy of users. This paper provides a comprehensive review of the current state of leveraging deep learning within edge computing to unleash the potential of IoT big data generated from IoT end devices. We believe that the revision will have a contribution to the development of DNA-based intelligent services and applications. It describes the different distributed training and inference architectures of deep learning models across multiple nodes of the edge computing platform. It also provides the different privacy-preserving approaches of deep learning on the edge computing environment and the various application domains where deep learning on the network edge can be useful. Finally, it discusses open issues and challenges leveraging deep learning within edge computing.
Vulnerability to Poverty and Its Determinants in Rural Ethiopia
Abrham Seyoum TSEHAY 연세대학교 빈곤문제국제개발연구원 2017 Journal of Poverty Alleviation and International D Vol.8 No.2
This study analyzes the vulnerability to poverty of smallholders in rural Ethiopia using a unique panel dataset, the Ethiopian Rural Household Survey which was collected from 15 peasant Associations covering 1359 households for the years 1994, 1999, 2004 and 2009. Three steps feasible generalized least squares and Principal component approaches are employed for this purpose. The results indicate that smallholders in rural Ethiopia in general are subjected to high levels of vulnerability to poverty as measured by a poverty threshold of 1 USD and a vulnerability threshold of 0. 5 using the consumption based approach. Both poverty incidence and vulnerability to poverty in rural Ethiopia are substantial but have opposite trends across survey rounds. While vulnerability to poverty increases steadily till 2004 before it moderately declines in the last round, the rate of poverty, on the contrary, declined consistently till the third round but considerably increased in the last round. However, vulnerability to poverty prevailed over poverty incidence in all the survey rounds indicating the need to give more focus on precautionary measures than merely safety net programs.
Yemane Teklay Seyoum,Syed Maaz Shahid,Sungoh Kwon 한국통신학회 2022 한국통신학회 학술대회논문집 Vol.2022 No.2
This paper reviews Multi-access Edge Computing (MEC) with an emphasis on major research areas, technology trends, and future research directions. MEC is envisioned to play a key role in meeting the unprecedented requirements of next-generation networks (5G, 6G and beyond) by providing computing and caching capabilities in close proximity of the users. It promises dramatic reduction in latency and mobile energy consumption, tackling the key challenges for realizing the Internet of Inteligent Things (IoIT) vision. The promised gains of MEC have inspired both academia and industry to conduct extensive research on developing the technology. However, despite the extensive research efforts in recent years, the field is still in its infancy. After conducting a comprehensive survey, we have identified computation offloading, caching, network automation, and edge artificial intelligence as major MEC research topics. We also presented research gaps pertaining to these key focus areas and recommended the investigation and exploitation of game theoretic strategies, big data analytics, distributed learning/federated learning, and programmable networks to tackle them.
Solomon Wagaw,Seyoum Mengistou,Abebe Getahun 한국수산과학회 2022 Fisheries and Aquatic Sciences Vol.25 No.7
Morphometric relationships, condition factor (CF) and reproductive biology are significant tools in fish stock management, fish biology, physiology, conservation and ecology. Growth and reproductive strategy of Oreochromis niloticus were studied from 343 fish specimens collected from Lake Shala between January and December 2018. Fish samples ranged from 7.7 cm to 33.0 cm in total length (TL) and from 7.80 g to 708.21 g in total weight (TW) were collected using gillnets of 4, 6, 8, 10, and 12 mm mesh sizes. The length-weight relationship of O. niloticus was TW = 0.0104TL3.19, indicating positive allometric growth of the fish. The sex ratio (0.93:1) was insignificant from the ideal fish distribution of 1:1 (χ2 = 0.47, p > 0.05). Mean CF for males, females and combined sexes was 1.04, 1.06 and 1.05, respectively and statistically insignificant (p > 0.05). The spawning peak occurred in July (rainy) and February (dry) periods, as defined by ripe females and the breeding season. Absolute mean fecundity was 806 eggs and correlated positively with TL and TW of the fish (p < 0.05) (F = 0.56TL2.29, R2 = 0.93, p < 0.05; F = 18.83TW0.67, R2 = 0.90, p < 0.05). The study provides the first detailed account of the morphometric relationships and reproductive biology of O. niloticus in Lake Shala, which can be used as baseline information for successive biological-based studies in Soda Lakes of Ethiopia.
Kejela Segni,Seyoum Nebyou 대한외상학회 2022 大韓外傷學會誌 Vol.35 No.3
Purpose: There is a strong correlation between trauma and pain. Pain increases the rate of depression, posttraumatic stress disorder, and even mortality in trauma patients. Methods: This institution-based, provider-blinded and patient-blinded, observational study was conducted among trauma patients treated at a specialized center . Over the course of 3 months, this study included patients who had no prior pain management at other hospitals before presentation, and who presented within 24 hours of the traumatic event. Results: Of the 74 patients evaluated, none of the patients had their pain level scored. The researcher-provided pain scale showed a severe subjective pain score for 79.7% of the patients and a severe functional activity score for 59.5% of the patients. Analgesia was provided at an average of 55.4 minutes after presentation and all patients received either diclofenac or tramadol. Satisfactory pain reduction after analgesia was 28.8% for patients initially complaining of severe pain, 54.6% for moderate pain, and 66.7% for mild pain, with the difference being statistically significant (P<0.05). Forty percent of patients discharged home received no analgesia after the first dose provided upon presentation. Conclusions: Pain scoring was nonexistent during the course of the study. The poor utilization rate of analgesia combination and opioids led to unsatisfactory pain outcomes in patients evaluated and followed for 24 hours after presentation.