• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.8C
• /
• pp.777-785
• /
• 2006

H.264 is the prominent video coding standard in various applications such as real-time video streaming and digital multimedia broadcasting, since it provides enhanced compression performance, error resilience tools, and network adaptation. Since compressed video stream is vulnerable to packet loss, error resilience and error concealment(EC) tools are essential for the transmission of video over the Internet. In this paper, we first propose a simple temporal EC method that improves the EC performance for intra-frame in H.264 when the amount of motion is relatively small. Then we propose a new hybrid EC method for intra-frame in H.264, which combines the spatial EC and temporal EC adaptively. The simulations are performed in packet-lossy environments, and the proposed hybrid EC method shows about 0.5-4dB PSNR improvement compared to the conventional spatial EC method that is used for intra-frame in H.264.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.2
• /
• pp.751-770
• /
• 2019

Action recognition is an essential task in computer vision due to the variety of prospective applications, such as security surveillance, machine learning, and human-computer interaction. The availability of more video data than ever before and the lofty performance of deep convolutional neural networks also make it essential for action recognition in video. Unfortunately, limited crafted video features and the scarcity of benchmark datasets make it challenging to address the multi-person action recognition task in video data. In this work, we propose a deep convolutional neural network-based Effective Hybrid Learning (EHL) framework for two-person interaction classification in video data. Our approach exploits a pre-trained network model (the VGG16 from the University of Oxford Visual Geometry Group) and extends the Faster R-CNN (region-based convolutional neural network a state-of-the-art detector for image classification). We broaden a semi-supervised learning method combined with an active learning method to improve overall performance. Numerous types of two-person interactions exist in the real world, which makes this a challenging task. In our experiment, we consider a limited number of actions, such as hugging, fighting, linking arms, talking, and kidnapping in two environment such simple and complex. We show that our trained model with an active semi-supervised learning architecture gradually improves the performance. In a simple environment using an Intelligent Technology Laboratory (ITLab) dataset from Inha University, performance increased to 95.6% accuracy, and in a complex environment, performance reached 81% accuracy. Our method reduces data-labeling time, compared to supervised learning methods, for the ITLab dataset. We also conduct extensive experiment on Human Action Recognition benchmarks such as UT-Interaction dataset, HMDB51 dataset and obtain better performance than state-of-the-art approaches.

• Smart Media Journal
• /
• v.1 no.1
• /
• pp.53-57
• /
• 2012

Smart phone and its applications are currently bringing about significant changes in our lives and it is expected that applying such technology in the area of agriculture could increase the value added and productivity of agriculture with its various uses. This paper proposed a smartphone-based application for monitoring and managing livestock barn in real-time anytime, anywhere. In the proposed application, the livestock barn environment and video information collected in ubiquitous livestock barn based on wireless sensor networks can be used by user to monitor the livestock barn in real-time through the use of smart phone to control the livestock barn facilities anytime, anywhere. This application can provide user convenience and increase productivity by allowing users to control their livestock barn facilities.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.8 no.2
• /
• pp.240-244
• /
• 2007

Non-intrusive methods based on active remote IR illumination fur eye tracking is important for many applications of vision-based man-machine interaction. One problem that has plagued those methods is their sensitivity to lighting condition change. This tends to significantly limit their scope of application. In this paper, we present a new real-time eye detection and tracking methodology that works under variable and realistic lighting conditions. Based on combining the bright-pupil effect resulted from IR light and the conventional appearance-based object recognition technique, our method can robustly track eyes when the pupils are not very bright due to significant external illumination interferences. The appearance model is incorporated in both eyes detection and tracking via the use of support vector machine and the mean shift tracking. Additional improvement is achieved from modifying the image acquisition apparatus including the illuminator and the camera.

• Journal of Broadcast Engineering
• /
• v.21 no.5
• /
• pp.782-791
• /
• 2016

The object segmentation algorithm from the background could be used for object recognition and tracking, and many applications. To segment objects, this paper proposes a method that refer to several initial frames with real-time processing at fixed camera. First we suggest the probability model to segment object and background and we enhance the performance of algorithm analyzing the color consistency and focus characteristic of camera for several initial frames. We compensate the segmentation result by using human skeleton characteristic among extracted objects. Last the proposed method has the applicability for various mobile application as we minimize computing complexity for real-time video processing.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2001.05a
• /
• pp.378-382
• /
• 2001

There are some technical problems in current internet-broadcasting. The load of server rises in proportion to user connection. And the inefficient usage of network bandwidth deteriorates quality of services and doesn't transport multimedia data in real time. To overcome above problems, multicast transport technology is applied to real-time multimedia data transport. But the reliability problem is still remained. This paper provides an efficient design of internet broadcasting server. We propose an multimedia data transport algorithm using adaptive encoding by grouping users according to similar connection environment. Performance evaluations show that the mechanism decreases the load of server and improves the quality of services.

• Journal of the Korea Society of Computer and Information
• /
• v.4 no.4
• /
• pp.155-162
• /
• 1999

Various applications of video conferencing are required real-time transmission in order to offer service of best effort in internet. Because the bandwidth of internet changes dynamically, appropriated QoS could not be guaranteed To resolve the problem. available bandwidth between sender and receiver is measured. And according to measured bandwidth, the transmission of multimedia data is controlled In this paper, we propose algorithm of efficient transmission for best QoS in internet According to a present status of network, we measure available bandwidth using feedback RTCP information and change a compression rate to reduce a producing CODEC data. And according to the priority that is measured by packet loss for received RTCP information, we abandon frames indicated as lower weight in transmission buffer of sender.

• ETRI Journal
• /
• v.33 no.3
• /
• pp.382-392
• /
• 2011

This paper proposes the streaming radio frequency identification (RFID) protocol to support robust data streaming in a passive communication, which is extended from the ISO18000-6 Type C RFID standard. By observing and modeling the unique bit error behavior through detailed analysis in this paper, we found that performance is significantly limited by inaccurate and unstable link frequencies as well as low SNR which are inevitable for passive devices. Based on the analysis, we propose a simple and efficient protocol to adaptively insert extra error control sequences in a packet for tolerating tough link condition while maximizing the throughput and preserving the minimal implementation cost. To evaluate effectiveness of our proposal in real-time streaming applications, we experimented on real-time H.264 video streaming and prototyped the system on FPGA. To our best knowledge, our paper is the first work to take analytical approach for maximizing the throughput and demonstrate the possibility of the realtime multimedia streaming transmission in the passive RFID system.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.9 no.4
• /
• pp.237-243
• /
• 2014

In this paper we present an real-time obstacle avoidance technique of autonomous mobile robot with steering system and implementation of user interface for mobile devices with Android platform. The direction of autonomous robot is determined by virtual force field concept, which is based on the distance information acquired from 5 ultrasonic sensors. It is converted to virtual repulsive force around the autonomous robot which is inversely proportional to the distance. The steering system with PD(proportional and derivative) controller moves the mobile robot to the determined target direction. We also use PSD(position sensitive detector) sensors to supplement ultrasonic sensors around dead angle area. The mobile robot communicates with Android mobile device and PC via Ethernet. The video information from CMOS camera mounted on the mobile robot is transmitted to Android mobile device and PC. And the user can control the mobile robot manually by transmitting commands on the user interface to it via Ethernet.

• Journal of Korea Multimedia Society
• /
• v.20 no.1
• /
• pp.17-31
• /
• 2017

The aim of the paper is to present an object recognition method toward augmented reality system that utilizes existing education instruments that was designed without any consideration on image processing and recognition. The light reflection, sizes, shapes, and color range of the existing target education instruments are major hurdles to our object recognition. In addition, the real-time performance requirements on embedded devices and user experience constraints for children users are quite challenging issues to be solved for our image processing and object recognition approach. In order to meet these requirements we employed a method cascading light-weight weak classification methods that are complimentary each other to make a resultant complicated and highly accurate object classifier toward practically reasonable precision ratio. We implemented the proposed method and tested the performance by video with more than 11,700 frames of actual playing scenario. The experimental result showed 0.54% miss ratio and 1.35% false hit ratio.