• Journal of Korea Multimedia Society
• /
• v.23 no.11
• /
• pp.1349-1360
• /
• 2020

In recent years, Convolutional Neural Networks (CNNs) have been successfully implemented in different tasks of computer vision. Since CNN models are the representatives of supervised learning algorithms, they demand large amount of data in order to train the classifiers. Thus, obtaining data with correct labels is imperative to attain the state-of-the-art performance of the CNN models. However, labelling datasets is quite tedious and expensive process, therefore real-life datasets often exhibit incorrect labels. Although the issue of poorly labelled datasets has been studied before, we have noticed that the methods are very complex and hard to reproduce. Therefore, in this research work, we propose Deep CleanNet - a considerably simple system that achieves competitive results when compared to the existing methods. We use K-means clustering algorithm for selecting data with correct labels and train the new dataset using a deep CNN model. The technique achieves competitive results in both training and validation stages. We conducted experiments using MNIST database of handwritten digits with 50% corrupted labels and achieved up to 10 and 20% increase in training and validation sets accuracy scores, respectively.

• Applied Microscopy
• /
• v.19 no.2
• /
• pp.74-84
• /
• 1989

Buffer soluble storage proteins of ginseng seed have been localized by electron microscopy using post-embedding immunocytochemical gold labelling technique. Major components of the storage proteins were revealed to be storage protein-1($SP_{1}$, MW 160,000) and storage protein-2($SP_{2}$, MW 70,000). Both of the storage proteins are glycoproteins. Anti-$SP_{1}$ and anti-$SP_{2}$ from rabbit, against $SP_1$ and $SP_2$, respectively, reacted on sections of ginseng endosperm tissue embedded in Spurr's epoxy resin. The rabbit antibodies were visualized indirectly by reaction with protein A labelled with colloidal gold. Both storage proteins were found to be accumulated together in the same protein bodies, but their relative contents are not equal.

• Journal of the Korean Institute of Electrical and Electronic Material Engineers
• /
• v.33 no.3
• /
• pp.239-244
• /
• 2020

This study examines the feasibility of the image deep learning method using convolution neural networks (CNNs) to maintain a porcelain insulator. Data augmentation is performed to prevent over-fitting, and the classification performance is evaluated by training the age, material, region, and pollution level of the insulator using image data in which the background and labelling are removed. Based on the results, it was difficult to predict the age, but it was possible to classify 76% of the materials, 60% of the pollution level, and more than 90% of the regions. From the results of this study, we identified the potential and limitations of the CNN classification for the four groups currently classified. However, it was possible to detect discoloration of the porcelain insulator resulting from physical, chemical, and climatic factors. Based on this, it will be possible to estimate the corrosion of the cap and discoloration of the porcelain caused by environmental deterioration, abnormal voltage, and lightning.

• Land and Housing Review
• /
• v.14 no.4
• /
• pp.95-102
• /
• 2023

Numerous investigations have explored the integration of machine leaning algorithms with high-resolution drone image for object detection in urban settings. However, a prevalent limitation in vehicle extraction studies involves the reliance on bounding boxes rather than instance segmentation. This limitation hinders the precise determination of vehicle direction and exact boundaries. Instance segmentation, while providing detailed object boundaries, necessitates labour intensive labelling for individual objects, prompting the need for research on automating unsupervised instance segmentation in vehicle extraction. In this study, a novel approach was proposed for vehicle extraction utilizing unsupervised SVM classification applied to vehicle bounding boxes in drone images. The method aims to address the challenges associated with bounding box-based approaches and provide a more accurate representation of vehicle boundaries. The study showed promising results, demonstrating an 89% accuracy in vehicle extraction. Notably, the proposed technique proved effective even when dealing with significant variations in spectral characteristics within the vehicles. This research contributes to advancing the field by offering a viable solution for automatic and unsupervised instance segmentation in the context of vehicle extraction from image.

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.11
• /
• pp.789-794
• /
• 2013

A vehicle tracking system makes it possible to induce the vehicle movement path for avoiding traffic congestion and to prevent traffic accidents in advance by recognizing traffic flow, monitoring vehicles, and detecting road accidents. To track the vehicles effectively, those which appear in a sequence of video frames need to identified by extracting the features of each object in the frames. Next, the identical vehicles over the continuous frames need to be recognized through the matching among the objects' feature values. In this paper, we identify objects by binarizing the difference image between a target and a referential image, and the labelling technique. As feature values, we use the center coordinate of the minimum bounding rectangle(MBR) of the identified object and the averages of 1D FFT(fast Fourier transform) coefficients with respect to the horizontal and vertical direction of the MBR. A vehicle is tracked in such a way that the pair of objects that have the highest similarity among objects in two continuous images are regarded as an identical object. The experimental result shows that the proposed method outperforms the existing methods that use geometrical features in tracking accuracy.

• Biomedical Science Letters
• /
• v.3 no.2
• /
• pp.131-138
• /
• 1997

IgG-$^{188}$Re conjugate was prepared for the diagnosis of abscess. The IgG molecules reduced by 2-mercaptoethanol contained 1.5 free sulfhydryl groups per IgG molecule. The reduced IgG molecule was labelled with $^{188}$Re through chelate to 99％ of labelling yield. The radiochemical purity of IgG-$^{188}$Re conjugate was maintained at 90％ in the presence of human serum for 1 hour. The IgG-$^{188}$Re was intravenously administered into staphylococcal abscess-bearing rats and their biodistribution was monitored at 4 and 24 hours post injection. The IgG-$^{188}$Re conjugate was moderately localized in the abscess tissue. This result implies that the IgG-$^{188}$Re conjugate can be a tool for abscess diagnosis. This technique can be applied for the preparation of various monoclonal antibody labelled with $^{188}$Re.

• Parasites, Hosts and Diseases
• /
• v.31 no.1
• /
• pp.37-42
• /
• 1993

Surface proteins of Trichomonqs unginnlis (T vqsinalis) were analyzed to study the antigenic variation. The surface proteins of protozoa were labelled by N- hydrokysuccinimide-biotin (NHS-biotins, the NHS-biotin-labelled proteins were immunoprecipitated with rabbit antiserum to purifjr the antigenic fractions and analysed by SDS-PAGE plus electroblotting. The results obtained in this study were as follows; Biotinylated T. uaginalis-proteins obtained from intact cell and cells disrupted prior to labelling were detected by antibiotin-peroxidase in Western blots. Labelled proteins were immunoprecipitated by T. vcqinalis-immunized rabbit serum and the six bands with, the molecular weights of 46, 60, 68, 90, 130 and 220 kDa were identified as having antigenicity. T unginalis HY-1,HY-15 and ATCC 50148 were immunoprecipitated by immune rabbit serum after biotinylation and there were no difference from antigenic bands among these strains by this tehchnique. In conclusion with the results obtained in the present study, it was assumed that surface proteins of T vaqinclis were labelled by biotinylation and the six labelled bands at 46, 60, 68, 90, 130 and 220 kDa in their molecular weight were identified as having antigenicity by immunoprecipitation (IPI and this biotinylation-IP technique may be used for further study of surface antigen of T vaginalis.

• Korean Journal of Animal Reproduction
• /
• v.20 no.2
• /
• pp.119-126
• /
• 1996

This study was carried out to determine the optimal condition for successful and efficient c cryopreservation of zygotes, 1-cell embryos, using EFS40 which was 40% (v/v) ethylene glycol diluted in DPBS medium containing 30% Fic-oll (w/v) and 0.3 M sucrose. After mouse zygote produced by IVF was vitrified by two freezing methods, the post-warming survival rates of 1-cell zygotes were assessed as cleavage to the 2-cell stage and development into the hatching blastocysts at 5 day. In the one-step method, when embryos were directly exposed to the vitrification solution at 25$^{\circ}C$ for 1 min., survival and development rates of zygotes were 85.5% and 31.9% In the two-step method, embryos were equilibrated with a dilute 20% EG for 1, 3, 5 min. before 1 min. exposure to EFS40, re-spectively. However, the rates of development (17.7, 3.3, 0%) were lower than that of one-step method. The highest survival rate (95.9%) was obtained by one-step method which exposes embryos in EFS40 for 30 sec. In this condition, 63. 8% of cleaved 2-cell developed into hatching blastocysts. In the cell number of Total and ICM using differential labelling technique, there are no significant differences in the cell number of Total and ICM between blastocysts devel oped in vitrified-thawed embryos (63.2${\pm}$16.9, 1 13.5${\pm}$4.0) and control balstocysts (54.0${\pm}$15.2, 1 12.3${\pm}$4.6). Therefore, these results show that mouse zygotes can be successfully cryopreserved by a simple vitrification method although developmental rates of vitrified embryos were reduced. In conclusion, this proposed vitrifi cation procedures can be useful in the cryopreservation of mouse IVF zygotes.

• Journal of Korean Philosophical Society
• /
• v.145
• /
• pp.271-296
• /
• 2018

In this article, I intend to reveal critically both the intrinsic crux and main problems of the politics of hostile exclusion based on the effect of labeling which was designed precisely as an impure political technique and has been operated for too long in Korean society by the conservative ruling class that centered on various negative ideological labels like 'pro-North Korean leftist forces.' Firstly, what is called the 'conservative ruling class' in Korean society is in itself an antinationalistic and antidemocratic pro-Japanese dictatorship group. Secondly, the conservative ruling class as a pro-Japanese dictatorship group has utilized politically the labeling effect which regards antigovernment Korean members as pro-North Korean or rebellious persons. This group's hostile politics, based on the ideological labelling effect, deprives antigovernment persons and groups of the qualification of Korean citizenship, in order to hold and retain their supreme power in Korean society. Thirdly, the conservative ruling class has attempted to stigmatize the citizens who participate in a movement for democracy as a pro-North Korean leftist force, but such a politically impure manner is typically completely unjustified groundless labeling. Fourthly, the attempt to define the conservative ruling class as a pro-Japanese dictatorship force is normatively justified and resonably appraised insofar as such a definition has been proved to be worthy of confidence. Finally, the trial to consider Roh's regime and pro-Roh (pro-Moon) groups as a kind of Yeongnam hegemonism by several critical intellectuals and current politicians from Honam region is not only merely a groundless and unconvincing labelling, but also the failed outcome of the attempt to systemize logically their emotional antipathy and repulsion toward Roh and pro-Roh (pro-Moon) groups.

• Clinical and Experimental Reproductive Medicine
• /
• v.21 no.1
• /
• pp.1-11
• /
• 1994

The occurrence and time course of capacitation, acrosomal loss, and hyperactivated motility require quantitative definition in order to characterize fertile human sperm. Recently the method has been developed to estimate the quality of spermatozoa by using kinematic parameters such as curvilinear velocity(VCL), average path velocity(VAP), linearity(LIN), straightness(STR), amplitude of lateral head displacement(ALH), and beat cross frequence(BCF) from Computer Assisted Sperm Analysis (CASA). In this study, using the Hamilton Thorn motility analyzer HTM 2030(Hamilton Thorn Research, Beverly, MA), we attempted to identify the spermatozoa with hyperactivated motility (HA) objectively and to monitor hyperactivation of human spermatozoa during incubation in capacitating media and after treatment of calcium ionophore as compared with acrosome status. And we examined whether HA are related to the result of SPA. Semen samples obtained from 16 healthy men were prepared by swim up technique and preincubated in a capacitating media(modified BWW medium) for 5 hours and treated with calcium ionophore solution. The acrosome reaction was detected with PSA-FITC labelling of the acrosome and in vitro sperm ferilizing capacity was assessed by the zona free hamster ovum penetration assay (SPA). The incidence of hyperactivated sperm was 2.6% in fresh semen, 14.3% of the swim up population, 13.7% after 5h of incubation. Significant increase of percentage of hyperactivated sperm was observed after the incubation (p<0.05) but after treatment, no significant changes of percentage of hyperactivated sperm(l1.8%) in contrast to significant rise in the percentage of acrosome reacted cells. Correlation analysis failed to show any significant relationship between the percentage of sperm with HA and SPA score. In conclusion, although no direct correlations were found between the results of SPA and HA, hyperactivation of sperm is associated with capacitation and monitoring hyperactivated sperm will be expected as a method of evaluating the functional quality of sperm such as SPA.