http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
내삼리 땅밀림형 산사태의 발생특성에 관한 분석(2) -발생원인을 중심으로-
박재현 ( Bak Jae Hyeon ),최경 ( Choe Gyeong ),마호섭 ( Ma Ho Seob ),이종학 ( Lee Jong Hag ),우보명 ( U Bo Myeong ) 한국산림과학회 2003 한국산림과학회지 Vol.92 No.6
This study was carried out to analyze rainfall events, geo-topographical characteristics at the landsliding area of about 3. lha in Nasamri, Juchonmyeon, Gimhaesi, Gyeongsangnam-do. There were many cracks and soil burying. Most rainfall was infiltrated from the surface layer to the subsurface layer. We found that subsurface water in the area was sprang out to the surface areas in the bottom of landsliding area. Heavily continued rainfall above 350㎜ which led to rapid shift of soil mass was one of the dominant factors in this landslide. We conclude that the occurrence of landslide was closely related to geo-topographical characteristics, hydrological conditions and rainfall mass events.
이철희(Cheol-Hee Lee),정윤주(Yoon-Ju Jeong),김태호(Taeho Kim),박재현(Jae-Hyeon Park),박성빈(Seongbin Bak),정철의(Chuleui Jung) 한국양봉학회 2019 韓國養蜂學會誌 Vol.34 No.3
One of the serious factors for honeybee decline is due to the various attacks from Vespa hornets, indigenous and invaded. Population monitoring as well as the alerting systems is requested against the Vespa. Automated image recognition is the primary step for the unmanned autonomous monitoring system development. This study compared the recent deep convolutional neural network (DCNN) algorithms such as AlexNet, VGG19, GoogLeNet, and ResNet50 for the best model selection for classification of 3 Vespa species, V. mandarinia, V. crabro and V. velutina. To evaluate classification performance, accuracy was utilized after transfer learning on each DCNN. As a result, the ResNet50 showed the best in terms of accuracy after sufficient training of 100 epochs. If performance and speed are considered simultaneously, AlexNet could be the alternative. The real-time monitoring system for objects requires both localization and classification. And Vespa occurrence or population change would need rapid recognition for the objects. Therefore speedy image recognition based on the DCNN, which combines localization and classification for objects in an image, should be considered in the future works.