• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.67 no.7
• /
• pp.865-870
• /
• 2018

A machine vision based industrial inspection includes defects detection and classification. Fast inspection is a fundamental problem for many applications of real-time vision systems. It requires little computation time and localizing defects robustly with high accuracy. Deep learning technique have been known not to be suitable for real-time applications. Recently a couple of fast region-based CNN algorithms for object detection are introduced, such as Faster R-CNN, and YOLOv2. We apply these methods for an industrial inspection problem. Three CNN based detection algorithms, VOV based CNN, Faster R-CNN, and YOLOv2, are experimented for defect detection on metal surface. The results for inspection time and various performance indices are compared and analysed.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.2
• /
• pp.800-816
• /
• 2018

In recent years, video surveillance research has been able to recognize various behaviors of pedestrians and analyze the overall situation of objects by combining image analysis technology and deep learning method. Human Activity Recognition (HAR), which is important issue in video surveillance research, is a field to detect abnormal behavior of pedestrians in CCTV environment. In order to recognize human behavior, it is necessary to detect the human in the image and to estimate the pose from the detected human. In this paper, we propose a novel approach for 2D Human Pose Estimation based on object detection using RGB-D information. By adding depth information to the RGB information that has some limitation in detecting object due to lack of topological information, we can improve the detecting accuracy. Subsequently, the rescaled region of the detected object is applied to ConVol.utional Pose Machines (CPM) which is a sequential prediction structure based on ConVol.utional Neural Network. We utilize CPM to generate belief maps to predict the positions of keypoint representing human body parts and to estimate human pose by detecting 14 key body points. From the experimental results, we can prove that the proposed method detects target objects robustly in occlusion. It is also possible to perform 2D human pose estimation by providing an accurately detected region as an input of the CPM. As for the future work, we will estimate the 3D human pose by mapping the 2D coordinate information on the body part onto the 3D space. Consequently, we can provide useful human behavior information in the research of HAR.

• Journal of Korea Multimedia Society
• /
• v.12 no.12
• /
• pp.1745-1754
• /
• 2009

PTZ (Pan-Tilt-Zoom) camera based surveillance systems are enlarging their field of application due to their wide observable area. We aimed to detect both static and moving objects in automated working space by using a PTZ camera. For object detection we used background difference method because of the high quality segmentation. However, the method has a problem called 'hole' that is caused by non-continuous surveillance of the PTZ camera and its own characteristics. Moreover, the occlusion which occurs when the moving object overlaps with the static object should be solved for robust object detection. In this paper, we suggest a region-based technique for updating background images thereby overcoming the hole and occlusion problem. Through experiments with real scenes, it was verified that meaningful static and/or moving objects were detected very well.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.3
• /
• pp.687-696
• /
• 2014

This paper presents a semantic object extraction method using user's stroke input, color, and depth information. It is supposed that a semantically meaningful object is surrounded with a few strokes from a user, and has similar depths all over the object. In the proposed method, deciding the region of interest (ROI) is based on the stroke input, and the semantically meaningful object is extracted by using color and depth information. Specifically, the proposed method consists of two steps. The first step is over-segmentation inside the ROI using color and depth information. The second step is semantically meaningful object extraction where over-segmented regions are classified into the object region and the background region according to the depth of each region. In the over-segmentation step, we propose a new marker extraction method where there are two propositions, i.e. an adaptive thresholding scheme to maximize the number of the segmented regions and an adaptive weighting scheme for color and depth components in computation of the morphological gradients that is required in the marker extraction. In the semantically meaningful object extraction, we classify over-segmented regions into the object region and the background region in order of the boundary regions to the inner regions, the average depth of each region being compared to the average depth of all regions classified into the object region. In experimental results, we demonstrate that the proposed method yields reasonable object extraction results.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.39 no.3
• /
• pp.399-407
• /
• 2019

In Korea, the construction industry has been known to have the highest risk of safety accidents compared to other industries. Therefore, in order to improve safety in the construction industry, several researches have been carried out from the past. This study aims at improving safety of labors in construction site by constructing an effective automatic safety helmet detection system using object detection algorithm based on image data of construction field. Deep learning was conducted using Region-based Fully Convolutional Network (R-FCN) which is one of the object detection algorithms based on Convolutional Neural Network (CNN) with Transfer Learning technique. Learning was conducted with 1089 images including human and safety helmet collected from ImageNet and the mean Average Precision (mAP) of the human and the safety helmet was measured as 0.86 and 0.83, respectively.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.22 no.9
• /
• pp.2043-2052
• /
• 1997

The object-oriented coding technique which is one of the coding methods in very low bit rate environment is suitable for videophone image sequence. The selection of source model affect image analysis. In this paper, an image analysis method for the object-oriented coding is presented. The process is composed of changed region detection andmotion estimateion. First, we use the standard deviation of frame difference as thrreshold to extract themoving area. If thesum of gray values in mask is greater than the threshold, the center pixel of the mask is regarded as moving region. After moving is detected in changed region by edge operator, observation point is determined from moving region. The motion is estimated by 6-parameter mapping method with determined observation point. The experimantal resutls show that the proposed method can significantly improve the image quality.

• Korean Journal of Remote Sensing
• /
• v.37 no.5_1
• /
• pp.1083-1094
• /
• 2021

In order to monitor nuclear activity in inaccessible areas, it is necessary to establish a methodology to analyze changesin nuclear activity-related objects using high-resolution satellite images. However, traditional object detection and change detection techniques using satellite images have difficulties in applying detection results to various fields because effects of seasons and weather at the time of image acquisition. Therefore, in this paper, an object of interest was detected in a satellite image using a deep learning model, and object changes in the satellite image were analyzed based on object detection results. An initial training of the deep learning model was performed using an open dataset for object detection, and additional training dataset for the region of interest were generated and applied to transfer learning. After detecting objects by multitemporal and multisensory satellite images, we tried to detect changes in objects in the images by using them. In the experiments, it was confirmed that the object detection results of various satellite images can be directly used for change detection for nuclear activity-related monitoring in inaccessible areas.

• Proceedings of the Korea Contents Association Conference
• /
• 2009.05a
• /
• pp.889-894
• /
• 2009

Accompanied by the rapid development of Computer Vision, Visual surveillance has achieved great evolution with more and more complicated processing. However there are still many problems to be resolved for robust and reliable visual surveillance, and the cast shadow occurring in motion detection process is one of them. Shadow pixels are often misclassified as object pixels so that they cause errors in localization, segmentation, tracking and classification of objects. This paper proposes a novel cast shadow removal method. As opposed to previous conventional methods, which considers pixel properties like intensity properties, color distortion, HSV color system, and etc., the proposed method utilizes observations about edge patterns in the shadow region in the current frame and the corresponding region in the background scene, and applies Laplacian edge detector to the blob regions in the current frame and the background scene. Then, the product of the outcomes of application determines whether the blob pixels in the foreground mask comes from object blob regions or shadow regions. The proposed method is simple but turns out practically very effective for Gaussian Mixture Model, which is verified through experiments.

• The KIPS Transactions:PartB
• /
• v.19B no.2
• /
• pp.107-112
• /
• 2012

An active object tracking algorithm using Pan and Tilt camera based in the stepwise application of region and color information from realtime image sequences is proposed. To reduce environment noises in input sequences, Gaussian filtering is performed first. An image is divided into background and objects by using the adaptive Gaussian mixture model. Once the target object is detected, an initial search window close to an object region is set up and color information is extracted from the region. We track moving objects in realtime by using the CAMShift algorithm which enables to trace objects in active camera with the color information. The proper tracking is accomplished by controlling the amount of pan and tilt to be placed the center position of object into the middle of field of view. The experimental results show that the proposed method is more effective than the hand-operated window method.

• Proceedings of the KSRS Conference
• /
• v.2
• /
• pp.610-614
• /
• 2006

In this paper the advanced method of geodesic active contours was developed for the task of building detection from aerial and satellite images. Automatic extraction of man-made structures including buildings, building blocks or roads from remote sensing data is useful for land use mapping, scene understanding, robotic navigation, image retrieval, surveillance, emergency management procedures, cadastral etc. A level set method based on a region-driven segmentation model was implemented with which building boundaries were detected, through this curve propagation technique. The essence of this approach is to optimize the position and the geometric form of the curve by measuring information along that curve, and within the regions that compose the image partition. To this end, one can consider uniform intensities inside objects and the background. Thus, given an initial position of the curve, one can determine global, region-driven functions and provide a statistical description of the inside and outside object area. The calculus of variations and a gradient descent method was used to optimize the variational functional by an iterative steady state process. Experimental results demonstrate the potential of the proposed processing scheme.