http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Test-Time Neural Style Transfer Augmentation For Polyp Classification
Zineb Tissir,Sang-Woong Lee 한국차세대컴퓨팅학회 2022 한국차세대컴퓨팅학회 학술대회 Vol.2022 No.10
Data augmentation has been employed in neural networks for building robust models, not exclusively in the training phase but also in the testing stage, where the predictions of every transformed image are aggregated to a greater lustiness and upgraded accuracy. Furthermore, deep learning approaches applied in data augmentation, namely adversarial training, GANs, and Neural Style Transfer were applied while training the models, neither while testing them. In this work, we present a study of applying test-time Neural Style Transfer transformation in medical images as a method of augmentation in test time. Besides, we display the experiment's results of a classification task. Results reveal that the synthesized samples employed as modified images in the test time significantly improved the performance of the classification model.
대장 질환 이미지 합성을 위한 CycleGAN 의 가능성 조사
Zineb Tissir,Sang-Woong Lee 한국차세대컴퓨팅학회 2022 한국차세대컴퓨팅학회 학술대회 Vol.2022 No.05
One of the main challenges of advancing medical imaging research is its data's privacy and sensitivity; sharing and distributing medical information is limited due to privacy concerns and the possible exploitation of personal information. Generative adversarial networks have impressive results in synthesizing new datasets from natural images and translating image to image. In the case of CycleGAN construct samples are done by translating the image from one domain to another. We present a study of the application of CycleGAN in medical imaging by converting standard images to images with a disease. Consequently, we test the generated dataset in a classification task and compare it with the original one. Results reveal that the synthesized samples could replace the original dataset
골프공 궤적 추적을 위한 영상처리 기반의 자동화 시스템
Ranjai Baidya,Hyun-Cheol Park,Zineb Tissir,Sang-Woong Lee 한국차세대컴퓨팅학회 2021 한국차세대컴퓨팅학회 학술대회 Vol.2021 No.05
Tracking a golf ball to visualize its trajectory had previously been done with complex and expensive systems. Those systems are not in the reach of every people. Here we suggest a more straightforward approach for the same purpose but utilizing only golf shots videos. The proposed system uses image processing and a deep neural network model, YOLO (You Only Live Once) V3. To achieve the goal of tracking the golf ball, the YOLO V3 network and Hough transform localize the ball’s initial position and the frame differencing technique tracks the golf ball. Upon implementation of the method to multiple videos, an acceptable result was obtained.