• Title/Summary/Keyword: Segmentation model

Search Result 1,031, Processing Time 0.027 seconds

Adversarial Wall: Physical Adversarial Attack on Cityscape Pretrained Segmentation Model (도시 환경에서의 이미지 분할 모델 대상 적대적 물리 공격 기법)

  • Suryanto, Naufal;Larasati, Harashta Tatimma;Kim, Yongsu;Kim, Howon
    • Proceedings of the Korea Information Processing Society Conference
    • /
    • 2022.11a
    • /
    • pp.402-404
    • /
    • 2022
  • Recent research has shown that deep learning models are vulnerable to adversarial attacks not only in the digital but also in the physical domain. This becomes very critical for applications that have a very high safety concern, such as self-driving cars. In this study, we propose a physical adversarial attack technique for one of the common tasks in self-driving cars, namely segmentation of the urban scene. Our method can create a texture on a wall so that it can be misclassified as a road. The demonstration of the technique on a state-of-the-art cityscape pretrained model shows a fairly high success rate, which should raise awareness of more potential attacks in self-driving cars.

Refinement of Ground Truth Data for X-ray Coronary Artery Angiography (CAG) using Active Contour Model

  • Dongjin Han;Youngjoon Park
    • International journal of advanced smart convergence
    • /
    • v.12 no.4
    • /
    • pp.134-141
    • /
    • 2023
  • We present a novel method aimed at refining ground truth data through regularization and modification, particularly applicable when working with the original ground truth set. Enhancing the performance of deep neural networks is achieved by applying regularization techniques to the existing ground truth data. In many machine learning tasks requiring pixel-level segmentation sets, accurately delineating objects is vital. However, it proves challenging for thin and elongated objects such as blood vessels in X-ray coronary angiography, often resulting in inconsistent generation of ground truth data. This method involves an analysis of the quality of training set pairs - comprising images and ground truth data - to automatically regulate and modify the boundaries of ground truth segmentation. Employing the active contour model and a recursive ground truth generation approach results in stable and precisely defined boundary contours. Following the regularization and adjustment of the ground truth set, there is a substantial improvement in the performance of deep neural networks.

Robust Coronary Artery Segmentation in 2D X-ray Images using Local Patch-based Re-connection Methods (지역적 패치기반 보정기법을 활용한 2D X-ray 영상에서의 강인한 관상동맥 재연결 기법)

  • Han, Kyunghoon;Jeon, Byunghwan;Kim, Sekeun;Jang, Yeonggul;Jung, Sunghee;Shim, Hackjoon;Chang, Hyukjae
    • Journal of Broadcast Engineering
    • /
    • v.24 no.4
    • /
    • pp.592-601
    • /
    • 2019
  • For coronary procedures, X-ray angiogram images are useful for diagnosing and assisting procedures. It is challenging to accurately segment a coronary artery using only a single segmentation model in 2D X-ray images due to a complex structure of three-dimensional coronary artery, especially from phenomenon of vessels being broken in the middle or end of coronary artery. In order to solve these problems, the initial segmentation is performed using an existing single model, and the candidate regions for the sophisticate correction is estimated based on the initial segment, and the local patch-based correction is performed in the candidate regions. Through this research, not only the broken coronary arteries are re-connected, but also the distal part of coronary artery that is very thin is additionally correctly found. Further, the performance can be much improved by combining the proposed correction method with any existing coronary artery segmentation method. In this paper, the U-net, a fully convolutional network was chosen as a segmentation method and the proposed correction method was combined with U-net to demonstrate a significant improvement in performance through X-ray images from several patients.

Training a semantic segmentation model for cracks in the concrete lining of tunnel (터널 콘크리트 라이닝 균열 분석을 위한 의미론적 분할 모델 학습)

  • Ham, Sangwoo;Bae, Soohyeon;Kim, Hwiyoung;Lee, Impyeong;Lee, Gyu-Phil;Kim, Donggyou
    • Journal of Korean Tunnelling and Underground Space Association
    • /
    • v.23 no.6
    • /
    • pp.549-558
    • /
    • 2021
  • In order to keep infrastructures such as tunnels and underground facilities safe, cracks of concrete lining in tunnel should be detected by regular inspections. Since regular inspections are accomplished through manual efforts using maintenance lift vehicles, it brings about traffic jam, exposes works to dangerous circumstances, and deteriorates consistency of crack inspection data. This study aims to provide methodology to automatically extract cracks from tunnel concrete lining images generated by the existing tunnel image acquisition system. Specifically, we train a deep learning based semantic segmentation model with open dataset, and evaluate its performance with the dataset from the existing tunnel image acquisition system. In particular, we compare the model performance in case of using all of a public dataset, subset of the public dataset which are related to tunnel surfaces, and the tunnel-related subset with negative examples. As a result, the model trained using the tunnel-related subset with negative examples reached the best performance. In the future, we expect that this research can be used for planning efficient model training strategy for crack detection.

3D Environmental Walkthrough Using The Integration of Multiple Segmentation Based Environment Models (다중 분할 기반 환경 모델의 통합에 의한 3차원 환경 탐색)

  • Ryoo, Seung-Taek
    • The Journal of Korean Association of Computer Education
    • /
    • v.8 no.1
    • /
    • pp.105-115
    • /
    • 2005
  • An environment model that is constructed using a single image has the problem of a blurring effect caused by the fixed resolution, and the stretching effect of the 3D model caused when information that does not exist on the image occurs due to the occlusion. This paper introduces the registration and integration method using multiple images to resolve the above problem. This method can represent parallax effect and expand the environment model to represent wide range of environment. The segmentation-based environment modeling method using multiple images can build a detail model with optimal resolution.

  • PDF

Automated texture mapping for 3D modeling of objects with complex shapes --- a case study of archaeological ruins

  • Fujiwara, Hidetomo;Nakagawa, Masafumi;Shibasaki, Ryosuke
    • Proceedings of the KSRS Conference
    • /
    • 2003.11a
    • /
    • pp.1177-1179
    • /
    • 2003
  • Recently, the ground-based laser profiler is used for acquisition of 3D spatial information of a rchaeological objects. However, it is very difficult to measure complicated objects, because of a relatively low-resolution. On the other hand, texture mapping can be a solution to complement the low resolution, and to generate 3D model with higher fidelity. But, a huge cost is required for the construction of textured 3D model, because huge labor is demanded, and the work depends on editor's experiences and skills . Moreover, the accuracy of data would be lost during the editing works. In this research, using the laser profiler and a non-calibrated digital camera, a method is proposed for the automatic generation of 3D model by integrating these data. At first, region segmentation is applied to laser range data to extract geometric features of an object in the laser range data. Various information such as normal vectors of planes, distances from a sensor and a sun-direction are used in this processing. Next, an image segmentation is also applied to the digital camera images, which include the same object. Then, geometrical relations are determined by corresponding the features extracted in the laser range data and digital camera’ images. By projecting digital camera image onto the surface data reconstructed from laser range image, the 3D texture model was generated automatically.

  • PDF

Accurate Parked Vehicle Detection using GMM-based 3D Vehicle Model in Complex Urban Environments (가우시안 혼합모델 기반 3차원 차량 모델을 이용한 복잡한 도시환경에서의 정확한 주차 차량 검출 방법)

  • Cho, Younggun;Roh, Hyun Chul;Chung, Myung Jin
    • The Journal of Korea Robotics Society
    • /
    • v.10 no.1
    • /
    • pp.33-41
    • /
    • 2015
  • Recent developments in robotics and intelligent vehicle area, bring interests of people in an autonomous driving ability and advanced driving assistance system. Especially fully automatic parking ability is one of the key issues of intelligent vehicles, and accurate parked vehicles detection is essential for this issue. In previous researches, many types of sensors are used for detecting vehicles, 2D LiDAR is popular since it offers accurate range information without preprocessing. The L shape feature is most popular 2D feature for vehicle detection, however it has an ambiguity on different objects such as building, bushes and this occurs misdetection problem. Therefore we propose the accurate vehicle detection method by using a 3D complete vehicle model in 3D point clouds acquired from front inclined 2D LiDAR. The proposed method is decomposed into two steps: vehicle candidate extraction, vehicle detection. By combination of L shape feature and point clouds segmentation, we extract the objects which are highly related to vehicles and apply 3D model to detect vehicles accurately. The method guarantees high detection performance and gives plentiful information for autonomous parking. To evaluate the method, we use various parking situation in complex urban scene data. Experimental results shows the qualitative and quantitative performance efficiently.

An Analysis of Market Segmentation and the Competitive Structure of the Shoes Market in Korea (우리나라 제화시장의 시장세분화 및 경쟁구조 분석)

  • Shin, Joung-Won;Hwang, Sun-Jin;Lee, Yun-Kyung
    • Journal of the Korean Society of Costume
    • /
    • v.58 no.7
    • /
    • pp.92-103
    • /
    • 2008
  • The purpose of this study was to subdivide the shoes market in Korea and to evaluate the size and competitive strength of each segment. In order to implement the purpose of this study, the data of 300 respondents were analyzed using CBC(Choice-Based Conjoint measurement) and mixture model. The part-worth utilities were then used to predict the impact of price change on the choice probability using the legit model. As a result, the mixture model showed the optimal segments number and the shoes market in Korea was divided into 4 segments. Each segment was identified by distinctive characteristics such as brands, price and demand for comfortable shoes. Also, as a result of grasping the competitive structure and the competitive strength by sub-markets, one group was sensitive to price according to each competitive situation, whereby the choice probability was greatly influenced, and the other group on the contrary. This study made it clear that discrimination between brands whose profits Increase sharply if price is lowered and brands whose profits do not increase even if price is lowered can help brand managers with their decision-making on price lowering.

Segmentation of Computed Tomography using The Geometric Active Contour Model (기하학적 동적 외곽선 모델을 이용한 X-ray 단층촬영영상의 영상추출)

  • Jang, D.P.;Kim, S.I.
    • Proceedings of the KOSOMBE Conference
    • /
    • v.1997 no.11
    • /
    • pp.541-545
    • /
    • 1997
  • This paper presents a modified geometric active contour model or edge detection and segmentation of computed tomography(CT) scan images. The method is based on the level setup approach developed by Osher and Sethian and the modeling of propagation fronts with curvature dependent speeds by Malladi. Based on above algorithms, the geometric active contour is obtained through a particular level set of hypersurface lowing along its gradient force and curvature force. This technique retains the attractive feature which is topological and geometric flexibility of the contour in recovering objects with complex shapes and unknown topologies. But there are limitations in this algorithm which are being not able to separate the object with weak difference from neighbor object. So we use speed limitation filter to overcome those problems. We apply a 2D model to various synthetic cases and the three cases of real CT scan images in order to segment objects with complicated shapes and topologies. From the results, the presented model confirms that it attracts very naturally and efficiently to the desired feature of CT scan images.

  • PDF

Accuracy Assessment of Forest Degradation Detection in Semantic Segmentation based Deep Learning Models with Time-series Satellite Imagery

  • Woo-Dam Sim;Jung-Soo Lee
    • Journal of Forest and Environmental Science
    • /
    • v.40 no.1
    • /
    • pp.15-23
    • /
    • 2024
  • This research aimed to assess the possibility of detecting forest degradation using time-series satellite imagery and three different deep learning-based change detection techniques. The dataset used for the deep learning models was composed of two sets, one based on surface reflectance (SR) spectral information from satellite imagery, combined with Texture Information (GLCM; Gray-Level Co-occurrence Matrix) and terrain information. The deep learning models employed for land cover change detection included image differencing using the Unet semantic segmentation model, multi-encoder Unet model, and multi-encoder Unet++ model. The study found that there was no significant difference in accuracy between the deep learning models for forest degradation detection. Both training and validation accuracies were approx-imately 89% and 92%, respectively. Among the three deep learning models, the multi-encoder Unet model showed the most efficient analysis time and comparable accuracy. Moreover, models that incorporated both texture and gradient information in addition to spectral information were found to have a higher classification accuracy compared to models that used only spectral information. Overall, the accuracy of forest degradation extraction was outstanding, achieving 98%.