• Proceedings of the Korean Society of Computer Information Conference
• /
• 2021.01a
• /
• pp.277-278
• /
• 2021

최근 손동작 인식에 대한 연구들이 활발하다. 하지만 대부분 Depth 정보를 포함한3D 정보를 필요로 한다. 이는 기존 연구들이 Depth 카메라 없이는 동작하지 않는다는 한계점이 있다는 것을 의미한다. 본 프로젝트는 Depth 카메라를 사용하지 않고 2D 이미지에서 Hand Keypoint Detection을 통해 손동작 인식을 하는 방법론을 제안한다. 학습 데이터 셋으로 Facebook에서 제공하는 InterHand2.6M 데이터셋[1]을 사용한다. 제안 방법은 크게 두 단계로 진행된다. 첫째로, Object Detection으로 Hand Detection을 수행한다. 데이터 셋이 어두운 배경에서 촬영되어 실 사용 환경에서 Detection 성능이 나오지 않는 점을 해결하기 위한 이미지 합성 Augmentation 기법을 제안한다. 둘째로, Keypoint Detection으로 21개의 Hand Keypoint들을 얻는다. 실험을 통해 유의미한 벡터들을 생성한 뒤 Annoy (Approximate nearest neighbors Oh Yeah) Tree를 생성한다. 생성된 Annoy Tree들로 후처리 작업을 거친 뒤 최종 Pose Estimation을 완료한다. Annoy Tree를 사용한 Pose Estimation에서는 NN(Neural Network)을 사용한 것보다 빠르며 동등한 성능을 냈다.

• Journal of KIISE:Databases
• /
• v.29 no.1
• /
• pp.79-88
• /
• 2002

Data on the Internet are usually represented and transfered as XML. the XML data is represented as a tree and therefore, object repositories are well-suited to store and query them due to their modeling power. XML queries are represented as regular path expressions and evaluated by traversing each object of the tree in object repositories. Several indexes are proposed to fast evaluate regular path expressions. However, in some cases they may not cover all possible paths because they require a great amount of disk space. In order to efficiently evaluate the queries in such cases, we propose an optimized traversing which combines the signature method and block traversing. The signature approach shrink the search space by using the signature information attached to each object, which hints the existence of a certain label in the sub-tree. The block traversing reduces disk I/O by early evaluating the reachable objects in a page. We conducted diverse experiments to show that the hybrid approach achieves a better performance than the other naive ones.

• Smart Media Journal
• /
• v.11 no.4
• /
• pp.9-18
• /
• 2022

Recently, research and system using AI is rapidly increasing in various fields. Smart farm using artificial intelligence and information communication technology is also being studied in agriculture. In addition, data-based precision agriculture is being commercialized by convergence various advanced technology such as autonomous driving, satellites, and big data. In Korea, the number of commercialization cases of facility agriculture among smart agriculture is increasing. However, research and investment are being biased in the field of facility agriculture. The gap between research and investment in facility agriculture and open-air agriculture continues to increase. The fields of fruit trees and plant factories have low research and investment. There is a problem that the big data collection and utilization system is insufficient. In this paper, we are proposed the system for determining the fruit tree growth stage using a deep learning-based object detection model. The system was proposed as a hybrid app for use in agricultural sites. In addition, we are implemented an object detection function for the fruit tree growth stage determine.

• Proceedings of the Korean Information Science Society Conference
• /
• 2007.10a
• /
• pp.34-35
• /
• 2007

• Journal of KIISE
• /
• v.43 no.2
• /
• pp.152-162
• /
• 2016

Most traditional garbage collectors commonly use the language level metadata, which is designed for pointer type searching. However, because it is difficult to use this metadata in non-volatile memory allocation platforms, a new garbage collection technique is essential for non-volatile memory utilization. In this paper, we design new metadata for managing information regarding non-volatile memory allocation called "Allocation Tree". This metadata is comprised of tree data structure for fast information lookup and a node that holds an allocation address and an object ID pair in key-value form. The Garbage Collector starts collecting when there are insufficient non-volatile memory spaces, and it compares user data and the allocation tree for garbage detection. We develop this algorithm in a persistent heap based non-volatile memory allocation platform called "HEAPO" for demonstration.

• Journal of Digital Convergence
• /
• v.10 no.10
• /
• pp.215-220
• /
• 2012

This paper proposes a ontology of tree structure approach for adaptive object recognition in a situation-variant environment. In this paper, we introduce a new concept, ontology of tree structure ontology, for context sensitivity, as we found that many developed systems work in a context-invariant environment. Due to the effects of illumination on a supreme obstinate designing context-sensitive recognition system, we have focused on designing such a context-variant system using ontology of tree structure. Ontology can be defined as an explicit specification of conceptualization of a domain typically captured in an abstract model of how people think about things in the domain. People produce ontologies to understand and explain underlying principles and environmental factors. In this research, we have proposed context ontology, context modeling, context adaptation, and context categorization to design ontology of tree structure based on illumination criteria. After selecting the proper light-ontology domain, we benefit from selecting a set of actions that produces better performance on that domain. We have carried out extensive experiments on these concepts in the area of object recognition in a dynamic changing environment, and we have achieved enormous success, which will enable us to proceed on our basic concepts.

• Journal of KIISE:Databases
• /
• v.33 no.3
• /
• pp.283-298
• /
• 2006

The TPR-tree is the time-parameterized indexing scheme that supports the querying of the current and projected future positions of such moving objects by representing the locations of the objects with their coordinates and velocity vectors. If this index is, however, used in environments that directions and velocities of moving objects, such as vehicles, are very often changed, it increases the communication cost between the server and moving objects because moving objects report their position to the server frequently when the direction and the velocity exceed a threshold value. To preserve the communication cost regularly, there can be used a manner that moving objects report their position to the server periodically. However, the periodical position report also has a problem that lineal time functions of the TPR-tree do not guarantee the accuracy of the object's positions if moving objects change their direction and velocity between position reports. To solve this problem, we propose the query processing scheme and the data structure using road networks for predicting uncertainty positions of moving objects, which is reported to the server periodically. To reduce an uncertainty of the query region, the proposed scheme restricts moving directions of the object to directions of road network's segments. To remove an uncertainty of changing the velocity of objects, it puts a maximum speed of road network segments. Experimental results show that the proposed scheme improves the accuracy for predicting positions of moving objects than other schemes based on the TPR-tree.

• The KIPS Transactions:PartD
• /
• v.16D no.1
• /
• pp.1-10
• /
• 2009

One of major purposes of a RFID system is to track moving objects using tags attached to the objects. Because a tagged object has both location and time information expressed as the location of the reader, we can index the trajectory of the object like existing spatiotemporal objects. More efficient tracking may be possible if a spatiotemporal trajectory can be formed of a tag, but there has not been much research on tag trajectory indexes. A characteristic that distinguishes tags from existing spatiotemporal objects is that a tag creates a separate trajectory in each reader by entering and then leaving the reader. As a result, there is a trajectory interruption interval between readers, in which the tag cannot be located, and this makes it difficult to track the tag. In addition, the point tags that only enter and don't leave readers do not create trajectories, so cannot be tracked. To solve this problem, we propose a tag trajectory index called TR-tree (tag trajectory R-tree in RFID system) that can track a tag by combining separate trajectories among readers into one trajectory. The results show that TR-tree, which overcomes the trajectory interruption superior performance than TPIR-tree and R-tree.

• Journal of Sensor Science and Technology
• /
• v.23 no.6
• /
• pp.377-382
• /
• 2014

Detecting and segmenting diffuse targets in laser ranging data is a critical problem for tactical reconnaissance. In this study, we propose a new method that facilitates the characterization of diffuse irregularly shaped objects using "spin images," i.e., local 2D histograms of laser returns oriented in 3D space, and a clustering process. The proposed "cluster-based spin imaging" method resolves the problem of using standard spin images for diffuse targets and it eliminates much of the computational complexity that characterizes the production of conventional spin images. The direct processing of pre-segmented laser points, including internal points that penetrate through a diffuse object's topmost surfaces, avoids some of the requirements of the approach used at present for spin image generation, while it also greatly reduces the high computational time overheads incurred by searches to find correlated images. We employed 3D airborne range data over forested terrain to demonstrate the effectiveness of this method in discriminating the different geometric structures of individual tree clusters. Our experiments showed that cluster-based spin images have the potential to separate classes in terms of different ages and portions of tree crowns.

• Journal of KIISE:Computing Practices and Letters
• /
• v.8 no.3
• /
• pp.303-317
• /
• 2002

In this paper, we design and implement E-COIRS enabling users to query with concepts and image features used for further refining the concepts. For example, E-COIRS supports the query "retrieve images containing black home appliance to north of reception set. "The query includes two types of concepts: IS-A and composite. "home appliance"is an IS-A concept, and "reception set" is a composite concept. For evaluating such a query. E-COIRS includes three important components: a visual image indexer, thesauri and a query processor. Each pair of objects in an mage captured by the visual image indexer is converted into a triple. The triple consists of the two object identifiers (oids) and their spatial relationship. All the features of an object is referenced by its old. A composite concept is detected by the triple thesaurus and IS-A concept is recolonized by the fuzzy term thesaurus. The query processor obtains an image set by matching each triple in a user with an inverted file and CS-Tree. To support efficient storage use and fast retrieval on high-dimensional feature vectors, E-COIRS uses Cell-based Signature tree(CS-Tree). E-COIRS is a more advanced content-based image retrieval system than other systems which support only concepts or image features.