논문 상세보기

한국물환경학회> 한국물환경학회지> 딥러닝 사물 인식 알고리즘(YOLOv3)을 이용한 미세조류 인식 연구

KCI등재

딥러닝 사물 인식 알고리즘(YOLOv3)을 이용한 미세조류 인식 연구

Microalgae Detection Using a Deep Learning Object Detection Algorithm, YOLOv3

박정수 ( Jungsu Park ) , 백지원 ( Jiwon Baek ) , 유광태 ( Kwangtae You ) , 남승원 ( Seung Won Nam ) , 김종락 ( Jongrack Kim )
  • : 한국물환경학회
  • : 한국물환경학회지 37권4호
  • : 연속간행물
  • : 2021년 07월
  • : 275-285(11pages)
한국물환경학회지

DOI

10.15681/KSWE.2021.37.4.275


목차

1. Introduction
2. Materials and Methods
3. Results and Discussion
4. Conclusion
Acknowledgement
References

키워드 보기


초록 보기

Algal bloom is an important issue in maintaining the safety of the drinking water supply system. Fast detection and classification of algae images are essential for the management of algal blooms. Conventional visual identification using a microscope is a labor-intensive and time-consuming method that often requires several hours to several days in order to obtain analysis results from field water samples. In recent decades, various deep learning algorithms have been developed and widely used in object detection studies. YOLO is a state-of-the-art deep learning algorithm. In this study the third version of the YOLO algorithm, namely, YOLOv3, was used to develop an algae image detection model. YOLOv3 is one of the most representative one-stage object detection algorithms with faster inference time, which is an important benefit of YOLO. A total of 1,114 algae images for 30 genera collected by microscope were used to develop the YOLOv3 algae image detection model. The algae images were divided into four groups with five, 10, 20, and 30 genera for training and testing the model. The mean average precision (mAP) was 81, 70, 52, and 41 for data sets with five, 10, 20, and 30 genera, respectively. The precision was higher than 0.8 for all four image groups. These results show the practical applicability of the deep learning algorithm, YOLOv3, for algae image detection.

UCI(KEPA)

간행물정보

  • : 공학분야  > 환경공학
  • : KCI등재
  • :
  • : 격월
  • : 2289-0971
  • : 2289-098X
  • : 학술지
  • : 연속간행물
  • : 1985-2021
  • : 2360


저작권 안내

한국학술정보㈜의 모든 학술 자료는 각 학회 및 기관과 저작권 계약을 통해 제공하고 있습니다.

이에 본 자료를 상업적 이용, 무단 배포 등 불법적으로 이용할 시에는 저작권법 및 관계법령에 따른 책임을 질 수 있습니다.

37권4호(2021년 07월) 수록논문
최근 권호 논문
| | | |

KCI등재

1황구지천 유역의 오염부하 특성 및 지류 영향 평가

저자 : 임수진 ( Su-jin Lim ) , 임병란 ( Byung-Ran Lim ) , 이한샘 ( Han-saem Lee ) , 강주형 ( Joo-hyoung Kang ) , 안태웅 ( Tae-ung Ahn ) , 신현상 ( Hyun-sang Shin )

발행기관 : 한국물환경학회 간행물 : 한국물환경학회지 37권 4호 발행 연도 : 2021 페이지 : pp. 249-262 (14 pages)

다운로드

(기관인증 필요)

초록보기

This study investigated the pollution characteristics of the main pollution zone in the Hwangguji watershed and the influence of the tributary on the main stream. The characteristics of the main pollution zone, including, the water quality index (WQI), stream rating, load duration curve (LDC), delivery load density (DLD), and contribution of the tributary to the mainstream, were evaluated by time-series visual heatmap. The WQI of the mainstream of Hwangguji was lowered to the poor (IV) level from the inflow point of Suwon stream (SW) and the LDC excess rate in the T-P was higher than that of BOD5, especially for the wet season, suggesting that management of non-point source with T-P is preferred. The contribution (%) of the tributaries in the upstream section of Hwangguji watershed were BOD5 14.54%, TOC 15.67%, T-N 5.43%, and T-P 6.97%. In particular, the Suwon sewage treatment plant located in the mainstream showed a high contribution of BOD5 (64.40%) and T-P (53.54%), respectively, due to the high discharge rate (6.019 ㎥/sec). Meanwhile, Sammi and Gal stream have a large impact on the mainstream with high DLD and poor WQI. Thus, both streams were considered as pollution hot spots. These results provide useful basic data for preparing more effective water quality improvement and management plans in the watershed.

KCI등재

2경안천 유역 수질 및 이행평가 자료를 통한 임의적 오염총량관리제도 시행의 성과 분석

저자 : 이범연 ( Lee Bum-Yeon ) , 이창희 ( Lee Chang-hee )

발행기관 : 한국물환경학회 간행물 : 한국물환경학회지 37권 4호 발행 연도 : 2021 페이지 : pp. 263-274 (12 pages)

다운로드

(기관인증 필요)

초록보기

This study presents the achievements and limitations of the voluntary-based Total Maximum Daily Load (TMDL) through statistical analysis of water quality monitoring data and performance assessments of TMDL plans implemented in the Gyeongan watershed. The results clearly showed that responsible local governments complied the allocated TMDL and the designated water quality goals were successfully achieved in the required period. This was possible because the Ministry of Environment provided innovative incentives, such as, relaxations of the existing tight land-use regulations and full-scale financial aids for constructing and operating public treatment facilities to draw local government voluntary participation. However, a couple of problems which decreased the effectiveness and efficiency of the voluntary TMDL were identified. The different TMDL implementation schedules between upstream (Yongin) and downstream (Gwangju) governments caused delay in water quality improvement and exaggerated TMDL allocation to the local development which made excessive investment in the treatment facilities. Although it is not directly related to the voluntary scheme, technical methods for establishing and assessing the water quality goals should be improved so that the effects of flow conditions on water quality are properly assessed. We expect that results of this case study contribute to developing a more effective voluntary-based scheme for the implementation of the so-called 'tributary TMDL' in the future.

KCI등재

3딥러닝 사물 인식 알고리즘(YOLOv3)을 이용한 미세조류 인식 연구

저자 : 박정수 ( Jungsu Park ) , 백지원 ( Jiwon Baek ) , 유광태 ( Kwangtae You ) , 남승원 ( Seung Won Nam ) , 김종락 ( Jongrack Kim )

발행기관 : 한국물환경학회 간행물 : 한국물환경학회지 37권 4호 발행 연도 : 2021 페이지 : pp. 275-285 (11 pages)

다운로드

(기관인증 필요)

초록보기

Algal bloom is an important issue in maintaining the safety of the drinking water supply system. Fast detection and classification of algae images are essential for the management of algal blooms. Conventional visual identification using a microscope is a labor-intensive and time-consuming method that often requires several hours to several days in order to obtain analysis results from field water samples. In recent decades, various deep learning algorithms have been developed and widely used in object detection studies. YOLO is a state-of-the-art deep learning algorithm. In this study the third version of the YOLO algorithm, namely, YOLOv3, was used to develop an algae image detection model. YOLOv3 is one of the most representative one-stage object detection algorithms with faster inference time, which is an important benefit of YOLO. A total of 1,114 algae images for 30 genera collected by microscope were used to develop the YOLOv3 algae image detection model. The algae images were divided into four groups with five, 10, 20, and 30 genera for training and testing the model. The mean average precision (mAP) was 81, 70, 52, and 41 for data sets with five, 10, 20, and 30 genera, respectively. The precision was higher than 0.8 for all four image groups. These results show the practical applicability of the deep learning algorithm, YOLOv3, for algae image detection.

KCI등재

4고속도로 노면퇴적물의 특성 및 도로청소에 의한 입도별 제거효율 분석

저자 : 강희만 ( Heeman Kang ) , 김황희 ( Hwang Hee Kim ) , 전지홍 ( Ji-hong Jeon )

발행기관 : 한국물환경학회 간행물 : 한국물환경학회지 37권 4호 발행 연도 : 2021 페이지 : pp. 286-295 (10 pages)

다운로드

(기관인증 필요)

초록보기

The removal efficiency of road-deposited sediment (SDR) by road sweeping was analyzed by performing particle size analysis before and after road sweeping at four highways during May to December 2019. The SDR accounted for the largest proportion in the range of 250 to 850 ㎛ and the degree of its proportion had an effect on the particle size distribution curve. The particle size distribution of the collected sediments showed a similar distribution at all sites. Below 75 ㎛, the removal efficiency of SDR showed a constant value around 40%, but above 75 ㎛, it increased as the particle size increased. The removal efficiency was 82-90% (average 86%) for gravel, 66-93% (average 79%) for coarse sand, 35-92% (average 64%) for fine sand, 29-69% (average 44%) for very fine sand, 19-58% (average 40%) for silt loading, 10-59% (average 40%) for TSP, 13-57% (average 40%) for PM10, and 15-61% (average 38%) for PM2.5. SDR removal efficiency showed an average of 69% for the four highways. It was found that if the amount of SDR was less than 100 g/m2, it was affected by the road surface condition and had a large regional deviation. As such, the amount of SDR and the removal efficiency increased. The fine particles, which have relatively low removal efficiency, contained a large amount of pollutants, which is an important factor in water and air pollution. Therefore, various measures to improve the removal efficiency of fine particles in SDR by road sweeping are needed.

KCI등재

5부산지역 도시하천 표층 퇴적물 오염도 평가에 관한 연구

저자 : 곽진숙 ( Jin-suk Kwag ) , 손정원 ( Jung-won Son ) , 김주인 ( Chu-In Kim ) , 송복주 ( Bok-Joo Song )

발행기관 : 한국물환경학회 간행물 : 한국물환경학회지 37권 4호 발행 연도 : 2021 페이지 : pp. 296-305 (10 pages)

다운로드

(기관인증 필요)

초록보기

This work investigated heavy metal pollution in surface sediments of rivers in Busan, Korea. Surface sediments were analyzed in order to conduct contamination assessment of organic matter, nutrients, and heavy metal concentrations. Contamination assessment of heavy metals was conducted using geoaccumulation index (Igeo), pollution load index (PLI), and potential ecological risk index (RI). Accumulation of organic matter and nutrients were affected by water discharged from sewage treatment plant. The concentrations of organic matter and nutrients were found to be greater in points which were close to the sewage treatment plant more than points furthest. The concentrations of Pb, Zn, Cu, Cd, Hg, As, Cr, and Ni were found to be greater in surface sediment more than in the background. The mean concentrations of heavy metals were in the order of Zn (323.5 mg/kg) > Cu (70.5 mg/kg) > Pb (39.8 mg/kg) > Cr (33.4 mg/kg) > Ni (13.5 mg/kg) > As (9.4 mg/kg) > Cd (0.84 mg/kg) > Hg (0.092 mg/kg). The result of geoaccumulation indices indicated that Hg > Cr > Cu > Ni > Zn > As > Pb > Cd were found in order of severe contamination by heavy metals. From PLI and RI analysis, it was evident that the Suyeonggang 2 was the most contaminated river.

KCI등재

6가축분뇨 퇴비·액비의 비료성분 및 중금속 함량에 관한 연구

저자 : 안태웅 ( Ahn Taeung ) , 김동민 ( Kim Dongmin ) , 이흥수 ( Lee Heungsoo ) , 신현상 ( Shin Hyunsang ) , 정유진 ( Chung Eugene )

발행기관 : 한국물환경학회 간행물 : 한국물환경학회지 37권 4호 발행 연도 : 2021 페이지 : pp. 306-314 (9 pages)

다운로드

(기관인증 필요)

초록보기

The application of organic fertilizer could be accompanied by potential hazards to soil and humans due to trace metals. Livestock manure compost·liquefied fertilizer is a well-established approach for the stabilization of nutrients and the reduction of pathogens and odors in manures, which can be evaluated as compost·liquefied. In this study, the livestock manure compost·liquefied fertilizers produced at 333 liquid manure public resource centers and liquid fertilizer distribution centers were collected from May to December 2019. The nutrient content (nitrogen, phosphorus, and potassium), physicochemical properties, and heavy metal content were investigated. The livestock manure compost·liquefied fertilizer was measured using a mechanical maturity measurement device. The organic matter, arsenic, cadmium, mercury, lead, chromium, copper, nickel, zinc, E. coli (O157:H7), Salmonella, etc. of the livestock manure compost·liquefied fertilizers were analyzed. The average heavy metal content in the livestock manure compost·liquefied fertilizer was as follows: Cr 2.9 mg/kg (0.2~8.7 mg/kg), Cu 20.4 mg/kg (1.6~74.1 mg/kg), Ni 1.3 mg/kg (0.4~4.2 mg/kg), and Zn 79.8 mg/kg (3.0~340.7 mg/kg). Although large-scale organic fertilizer plants and resources recycling centers produce good organic (liquid) fertilizers with proper components, it is necessary to standardize livestock manure compost·liquefied fertilizer in order to facilitate efforts to turn livestock manure into useful resources.

1
권호별 보기
같은 권호 다른 논문
| | | | 다운로드

KCI등재

1황구지천 유역의 오염부하 특성 및 지류 영향 평가

저자 : 임수진 ( Su-jin Lim ) , 임병란 ( Byung-Ran Lim ) , 이한샘 ( Han-saem Lee ) , 강주형 ( Joo-hyoung Kang ) , 안태웅 ( Tae-ung Ahn ) , 신현상 ( Hyun-sang Shin )

발행기관 : 한국물환경학회 간행물 : 한국물환경학회지 37권 4호 발행 연도 : 2021 페이지 : pp. 249-262 (14 pages)

다운로드

(기관인증 필요)

초록보기

This study investigated the pollution characteristics of the main pollution zone in the Hwangguji watershed and the influence of the tributary on the main stream. The characteristics of the main pollution zone, including, the water quality index (WQI), stream rating, load duration curve (LDC), delivery load density (DLD), and contribution of the tributary to the mainstream, were evaluated by time-series visual heatmap. The WQI of the mainstream of Hwangguji was lowered to the poor (IV) level from the inflow point of Suwon stream (SW) and the LDC excess rate in the T-P was higher than that of BOD5, especially for the wet season, suggesting that management of non-point source with T-P is preferred. The contribution (%) of the tributaries in the upstream section of Hwangguji watershed were BOD5 14.54%, TOC 15.67%, T-N 5.43%, and T-P 6.97%. In particular, the Suwon sewage treatment plant located in the mainstream showed a high contribution of BOD5 (64.40%) and T-P (53.54%), respectively, due to the high discharge rate (6.019 ㎥/sec). Meanwhile, Sammi and Gal stream have a large impact on the mainstream with high DLD and poor WQI. Thus, both streams were considered as pollution hot spots. These results provide useful basic data for preparing more effective water quality improvement and management plans in the watershed.

KCI등재

2경안천 유역 수질 및 이행평가 자료를 통한 임의적 오염총량관리제도 시행의 성과 분석

저자 : 이범연 ( Lee Bum-Yeon ) , 이창희 ( Lee Chang-hee )

발행기관 : 한국물환경학회 간행물 : 한국물환경학회지 37권 4호 발행 연도 : 2021 페이지 : pp. 263-274 (12 pages)

다운로드

(기관인증 필요)

초록보기

This study presents the achievements and limitations of the voluntary-based Total Maximum Daily Load (TMDL) through statistical analysis of water quality monitoring data and performance assessments of TMDL plans implemented in the Gyeongan watershed. The results clearly showed that responsible local governments complied the allocated TMDL and the designated water quality goals were successfully achieved in the required period. This was possible because the Ministry of Environment provided innovative incentives, such as, relaxations of the existing tight land-use regulations and full-scale financial aids for constructing and operating public treatment facilities to draw local government voluntary participation. However, a couple of problems which decreased the effectiveness and efficiency of the voluntary TMDL were identified. The different TMDL implementation schedules between upstream (Yongin) and downstream (Gwangju) governments caused delay in water quality improvement and exaggerated TMDL allocation to the local development which made excessive investment in the treatment facilities. Although it is not directly related to the voluntary scheme, technical methods for establishing and assessing the water quality goals should be improved so that the effects of flow conditions on water quality are properly assessed. We expect that results of this case study contribute to developing a more effective voluntary-based scheme for the implementation of the so-called 'tributary TMDL' in the future.

KCI등재

3딥러닝 사물 인식 알고리즘(YOLOv3)을 이용한 미세조류 인식 연구

저자 : 박정수 ( Jungsu Park ) , 백지원 ( Jiwon Baek ) , 유광태 ( Kwangtae You ) , 남승원 ( Seung Won Nam ) , 김종락 ( Jongrack Kim )

발행기관 : 한국물환경학회 간행물 : 한국물환경학회지 37권 4호 발행 연도 : 2021 페이지 : pp. 275-285 (11 pages)

다운로드

(기관인증 필요)

초록보기

Algal bloom is an important issue in maintaining the safety of the drinking water supply system. Fast detection and classification of algae images are essential for the management of algal blooms. Conventional visual identification using a microscope is a labor-intensive and time-consuming method that often requires several hours to several days in order to obtain analysis results from field water samples. In recent decades, various deep learning algorithms have been developed and widely used in object detection studies. YOLO is a state-of-the-art deep learning algorithm. In this study the third version of the YOLO algorithm, namely, YOLOv3, was used to develop an algae image detection model. YOLOv3 is one of the most representative one-stage object detection algorithms with faster inference time, which is an important benefit of YOLO. A total of 1,114 algae images for 30 genera collected by microscope were used to develop the YOLOv3 algae image detection model. The algae images were divided into four groups with five, 10, 20, and 30 genera for training and testing the model. The mean average precision (mAP) was 81, 70, 52, and 41 for data sets with five, 10, 20, and 30 genera, respectively. The precision was higher than 0.8 for all four image groups. These results show the practical applicability of the deep learning algorithm, YOLOv3, for algae image detection.

KCI등재

4고속도로 노면퇴적물의 특성 및 도로청소에 의한 입도별 제거효율 분석

저자 : 강희만 ( Heeman Kang ) , 김황희 ( Hwang Hee Kim ) , 전지홍 ( Ji-hong Jeon )

발행기관 : 한국물환경학회 간행물 : 한국물환경학회지 37권 4호 발행 연도 : 2021 페이지 : pp. 286-295 (10 pages)

다운로드

(기관인증 필요)

초록보기

The removal efficiency of road-deposited sediment (SDR) by road sweeping was analyzed by performing particle size analysis before and after road sweeping at four highways during May to December 2019. The SDR accounted for the largest proportion in the range of 250 to 850 ㎛ and the degree of its proportion had an effect on the particle size distribution curve. The particle size distribution of the collected sediments showed a similar distribution at all sites. Below 75 ㎛, the removal efficiency of SDR showed a constant value around 40%, but above 75 ㎛, it increased as the particle size increased. The removal efficiency was 82-90% (average 86%) for gravel, 66-93% (average 79%) for coarse sand, 35-92% (average 64%) for fine sand, 29-69% (average 44%) for very fine sand, 19-58% (average 40%) for silt loading, 10-59% (average 40%) for TSP, 13-57% (average 40%) for PM10, and 15-61% (average 38%) for PM2.5. SDR removal efficiency showed an average of 69% for the four highways. It was found that if the amount of SDR was less than 100 g/m2, it was affected by the road surface condition and had a large regional deviation. As such, the amount of SDR and the removal efficiency increased. The fine particles, which have relatively low removal efficiency, contained a large amount of pollutants, which is an important factor in water and air pollution. Therefore, various measures to improve the removal efficiency of fine particles in SDR by road sweeping are needed.

KCI등재

5부산지역 도시하천 표층 퇴적물 오염도 평가에 관한 연구

저자 : 곽진숙 ( Jin-suk Kwag ) , 손정원 ( Jung-won Son ) , 김주인 ( Chu-In Kim ) , 송복주 ( Bok-Joo Song )

발행기관 : 한국물환경학회 간행물 : 한국물환경학회지 37권 4호 발행 연도 : 2021 페이지 : pp. 296-305 (10 pages)

다운로드

(기관인증 필요)

초록보기

This work investigated heavy metal pollution in surface sediments of rivers in Busan, Korea. Surface sediments were analyzed in order to conduct contamination assessment of organic matter, nutrients, and heavy metal concentrations. Contamination assessment of heavy metals was conducted using geoaccumulation index (Igeo), pollution load index (PLI), and potential ecological risk index (RI). Accumulation of organic matter and nutrients were affected by water discharged from sewage treatment plant. The concentrations of organic matter and nutrients were found to be greater in points which were close to the sewage treatment plant more than points furthest. The concentrations of Pb, Zn, Cu, Cd, Hg, As, Cr, and Ni were found to be greater in surface sediment more than in the background. The mean concentrations of heavy metals were in the order of Zn (323.5 mg/kg) > Cu (70.5 mg/kg) > Pb (39.8 mg/kg) > Cr (33.4 mg/kg) > Ni (13.5 mg/kg) > As (9.4 mg/kg) > Cd (0.84 mg/kg) > Hg (0.092 mg/kg). The result of geoaccumulation indices indicated that Hg > Cr > Cu > Ni > Zn > As > Pb > Cd were found in order of severe contamination by heavy metals. From PLI and RI analysis, it was evident that the Suyeonggang 2 was the most contaminated river.

KCI등재

6가축분뇨 퇴비·액비의 비료성분 및 중금속 함량에 관한 연구

저자 : 안태웅 ( Ahn Taeung ) , 김동민 ( Kim Dongmin ) , 이흥수 ( Lee Heungsoo ) , 신현상 ( Shin Hyunsang ) , 정유진 ( Chung Eugene )

발행기관 : 한국물환경학회 간행물 : 한국물환경학회지 37권 4호 발행 연도 : 2021 페이지 : pp. 306-314 (9 pages)

다운로드

(기관인증 필요)

초록보기

The application of organic fertilizer could be accompanied by potential hazards to soil and humans due to trace metals. Livestock manure compost·liquefied fertilizer is a well-established approach for the stabilization of nutrients and the reduction of pathogens and odors in manures, which can be evaluated as compost·liquefied. In this study, the livestock manure compost·liquefied fertilizers produced at 333 liquid manure public resource centers and liquid fertilizer distribution centers were collected from May to December 2019. The nutrient content (nitrogen, phosphorus, and potassium), physicochemical properties, and heavy metal content were investigated. The livestock manure compost·liquefied fertilizer was measured using a mechanical maturity measurement device. The organic matter, arsenic, cadmium, mercury, lead, chromium, copper, nickel, zinc, E. coli (O157:H7), Salmonella, etc. of the livestock manure compost·liquefied fertilizers were analyzed. The average heavy metal content in the livestock manure compost·liquefied fertilizer was as follows: Cr 2.9 mg/kg (0.2~8.7 mg/kg), Cu 20.4 mg/kg (1.6~74.1 mg/kg), Ni 1.3 mg/kg (0.4~4.2 mg/kg), and Zn 79.8 mg/kg (3.0~340.7 mg/kg). Although large-scale organic fertilizer plants and resources recycling centers produce good organic (liquid) fertilizers with proper components, it is necessary to standardize livestock manure compost·liquefied fertilizer in order to facilitate efforts to turn livestock manure into useful resources.

1
발행기관 최신논문
자료제공: 네이버학술정보
발행기관 최신논문
자료제공: 네이버학술정보

내가 찾은 최근 검색어

최근 열람 자료

맞춤 논문

보관함

내 보관함
공유한 보관함

1:1문의

닫기