• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.1
• /
• pp.44-49
• /
• 2021

Recently, due to COVID-19, studies have been popularly worked to apply neural network to mask wearing automatic detection system. For applying neural networks, the 1-stage detection or 2-stage detection methods are used, and if data are not sufficiently collected, the pretrained neural network models are studied by applying fine-tuning techniques. In this paper, the system is consisted of 2-stage detection method that contain MTCNN model for face recognition and ResNet model for mask detection. The mask detector was experimented by applying five ResNet models to improve accuracy and fps in various environments. Training data used 17,217 images that collected using web crawler, and for inference, we used 1,913 images and two one-minute videos respectively. The experiment showed a high accuracy of 96.39% for images and 92.98% for video, and the speed of inference for video was 10.78fps.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.9
• /
• pp.157-167
• /
• 2022

This study attempted to investigate what kind of perception people in their 20s have about masks and to find out the characteristics of each type by categorizing the perception. The Q methodology was used for the study. The cognition types of masks were categorized into three. Type 1 was a 'always wear impact-important type' that always wears masks and thinks masks affect non-verbal communication and the wearer's image. Type 2 was a 'function-important negative recognition type' that wears masks to prevent germs and thinks that masks have a great negative impact. Type 3 was a 'concealment wear positive image type' that wears a mask to cover the face and thinks that a person looks young when wearing a mask. It is thought that the development of masks of various designs and functions reflecting the needs of consumers should be carried out. Also, it is thought that various products should be developed and sold so that consumers can choose according to important considerations such as design, fit, and function.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.413-414
• /
• 2024

영상 인식을 위한 학습 데이터 구성 단계에서 에지는 물체의 크기, 방향 등의 정보를 포함하고 있어 영상의 특징으로 사용한다. 본 논문에서는 얼굴 인식을 위하여 소벨 마스크를 사용하여 원영상과 압축영상 그리고 에지영상간의 학습에 따른 인식 정도를 파악하고자 한다. 실험결과, 원영상 그대로 인식하는 것보다 에지 영상에 의한 학습 속도에 차이가 있음을 알 수 있었다.

• Journal of the Korean Society of Fashion and Beauty
• /
• v.5 no.1 s.12
• /
• pp.97-111
• /
• 2007

This study was motivated not only by the curiosity of clown as a comedian, but also by the clown's special makeup coordination. To analyze the clown makeup coordination Carnivalism, the researcher inquired 1) clown's meaning, sorts, periodic roles and stages, 2) Mikhail Bakhtin's Carnival theory, and 3) makeup coordination of clowns (Harlequin, Pierrot, Auguste and White Face Clown) as the theoretical background. The clown makeup coordination carnival characteristics were shown in masquerade, amusement and ambiguity. The masquerade was expressed on the mask for Harlequin and white makeup for Pierrot, Auguste and White Face Clown. The amusement was revealed on the fall and the inharmony of the grotesque makeup and childish costumes. And also the ambiguity was expressed on the sexless and the infant of clown's real nature and costumes.

• The KIPS Transactions:PartB
• /
• v.8B no.1
• /
• pp.20-27
• /
• 2001

얼굴검출은 현재 많은 연구가 활발히 진행되고 있는 분야로 보안, 인식 등 다양한 응용분야를 갖는다. 본 논문은 카메라가 화자의 이동에 따라 이를 추적하여 회전하고 회의상황에 맞는 앵글을 유지하는 지능형 영상회의 시스템 개발의 기본요소인 화자검출의 선행단계로 얼굴검출에 대한 새로운 방법을 제안한다. RGB 색 공간의 입력영상을 YIQ 공간으로 변환한 후 IQ 성분은 피부영역검출에 Y 성분은 얼굴의 특성을 추출하는데 사용된다. 색 분포도를 이용하여 피부영역을 검출하고, 마스크를 누적 적용하여 잡음을 제거한 후 얼굴의 구조적인 특성과 명암의 분포를 이용하여 얼굴영역이 검출된다. 실험결과 다양한 배경의 영상에서 여러 명의 얼굴이 오류 없이 검출됨이 관찰되었다.

• Journal of KIISE:Software and Applications
• /
• v.36 no.11
• /
• pp.901-911
• /
• 2009

In this paper, we propose a construction technique of two-level composite gradient map to automatically track a face with large movement in successive frames. Our method is composed of three main steps. First, the gradient maps with two-level resolution are generated for fast convergence of active contour. Second, to recognize the variations of face between successive frames and remove the neighbor background, weighted composite gradient map is generated by combining the composite gradient map and difference mask of previous and current frames. Third, to prevent active contour from converging local minima, the energy slope is generated by using closing operation. In addition, the fast closing operation is proposed to accelerate the processing time of closing operation. For performance evaluation, we compare our method with previous active contour model-based face tracking methods using a visual inspection, robustness test and processing time. Experimental results show that our method can effectively track the face with large movement and robustly converge to the optimal position even in frames with complicated background.

• The Journal of the Korea Contents Association
• /
• v.14 no.11
• /
• pp.7-15
• /
• 2014

This paper proposed a recognition algorithm of human facial expressions using the PCA and the template matching. Firstly, face image is acquired using the Haar-like feature mask from an input image. The face image is divided into two images. One is the upper image including eye and eyebrow. The other is the lower image including mouth and jaw. The extraction of facial components, such as eye and mouth, begins getting eye image and mouth image. Then an eigenface is produced by the PCA training process with learning images. An eigeneye and an eigenmouth are produced from the eigenface. The eye image is obtained by the template matching the upper image with the eigeneye, and the mouth image is obtained by the template matching the lower image with the eigenmouth. The face recognition uses geometrical properties of the eye and mouth. The simulation results show that the proposed method has superior extraction ratio rather than previous results; the extraction ratio of mouth image is particularly reached to 99%. The face recognition system using the proposed method shows that recognition ratio is greater than 80% about three facial expressions, which are fright, being angered, happiness.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.9 no.3
• /
• pp.187-192
• /
• 2009

In this paper the extended algorithm of valley energy was proposed In order to detect the robust features of the face The proposed energy had the variable type without the constant size of valley mask for measuring the gray value among pixels By accumulating the results generated by the various masks the information of valley energy was so diversified. Then the robust energy which is independent of the environments was maded. The various experiments proved that The proposed method showed the detection rate of 98 percentage in the features of the face region.

• Maxillofacial Plastic and Reconstructive Surgery
• /
• v.31 no.1
• /
• pp.53-60
• /
• 2009

Maxillary deficiency, anterior cross bite, constriction of maxillary arch, malaligned teeth are frequently observed in patients with cleft lip and palate. Surgery and orthodontics, combined intervention are needed to correct maxillary deficiency. Distraction osteogenesis that currently used has many advantages like less relapse tendency, more advancement of maxilla, capable in growing patients. In case 1, 18 years old girl with BCLP had severe midfacial deficiency and multiple missing of teeth. LeFort I osteotomy, followed by maxillary distraction osteogenesis utilizing rigid external distraction device(RED) system, was performed. After a 6-day latency period, distraction proceeded at a rate of 1mm per day (at 1st week, 1.5mm/day). Total advancement was 19mm. The RED device left in place for the additional 4 weeks for consolidation. After the RED device was removed, face mask was applied with elastic traction for 5 weeks. After achieving acceptable facial appearance and occlusion, orthodontic appliance was removed. The results after 4 years follow-up was sustained pretty well without aggravation of velopharyngeal function. In case 2, 22 years old man with UCLP had severe midfacial deficiency and palatally erupted upper 2nd premolars due to arch length discrepancy, but the anterior segment of maxillary did not show constriction and crowding. patient had no arch width discrepancy, crowding was concentrated on premolar region. Segmental LeFort I osteotomy was performed. After a 6 - day latency period, using internal distraction device, distraction proceeded at a 0.5mm per day(at 1st week, 0.75 - 1 mm/day). Total advancement was 15mm. After internal distraction device was removed, face mask was applied with elastic traction for 4 weeks. After surgical-orthodontic treatment, facial appearance and occlusion was improved pretty good, and after 46 months follow-up the result was retained well.

• Journal of Korean Society of Occupational and Environmental Hygiene
• /
• v.6 no.2
• /
• pp.176-186
• /
• 1996

The fit of a respirator to the face of an individual can be determined by a qualitative fit test (QLFT) or a quantitative fit test (QNFT). The pass/fail decision from a QLFT or QNFT for the same respirator on the same individual may vary from one wearing to the next, because the human facial features are complex and the respirator may not fit to the face in the same way every time it is worn. This study reports how the fit factors (FF) resulting from a QNFT on an individual vary from day to day and depend on the movements in the six fit test exercises. The reported FFs provide an objective and numerical basis (FF) which does not depend on the subject's voluntary or involuntary response. Four half-mask (H1-H4) and four full-facepiece respirators (F1-F4) were fit tested on one wearer 10 times a day for 5 days with a PortaCount (model 8010, TSI). The FFs obtained for each set of 10 fit tests on a specific day and 50 fit tests on five days involving one of the six exercise regimes have been recorded as log-normal distributions. All of the geometric standard deviations (GSD) of the overall FFs varied widely among every wearing and day except for H1 and F3, and the variability of the half-mask respirators was larger than that of the full-facepiece respirators. Among the six exercise regimes, reading or talking (RT) had markedly the lowest exercise FFs on the tested individual. Generally, there were significant differences between the first normal breathing (NB1) FFs and the remaining exercise FFs.