Recently data-driven image inpainting methods have made inspiring progress, impacting fundamental editing tasks such as object removal and damaged repairing. These are more effective than classic approaches, however, due to memory limitations they can only handle low-resolution inputs, typically smaller 1K. Meanwhile, the resolution of photos captured with mobile devices increases up 8K. Naive up-sampling inpainted result merely yield a large yet blurry result. Whereas, adding high-frequency...

The knowledge graph completion (KGC) task aims to predict missing links in graphs. Recently, several KGC models based on translational distance or semantic matching methods have been proposed and achieved meaningful results. However, existing a significant shortcoming–they cannot train entity embedding when an does not appear the training phase. As result, such use randomly initialized embeddings for entities that are unseen phase cause critical decrease performance during test To solve this...

This paper presents a novel approach to real-time 3D modeling of workspace for manipulative robotic tasks. First, we establish the three fundamental principles that human uses and interacting with environment. These have led development an integrated modeling, as follows: 1) It starts rapid but approximate characterization geometric configuration by identifying global plane features. 2) quickly recognizes known objects in replaces them their models database based on in-situ registration. 3)...

This paper provides a novel approach for user oriented language model face detection. Even though there are many open source or commercial libraries to solve the problem of detection, they still hard use because require specific knowledge on details algorithmic techniques. proposes high-level detection with which users can develop systems easily and even without theories algorithms. Important conditions firstly considered categorize large space The identified here then represented as...

Recently data-driven image inpainting methods have made inspiring progress, impacting fundamental editing tasks such as object removal and damaged repairing. These are more effective than classic approaches, however, due to memory limitations they can only handle low-resolution inputs, typically smaller 1K. Meanwhile, the resolution of photos captured with mobile devices increases up 8K. Naive up-sampling inpainted result merely yield a large yet blurry result. Whereas, adding high-frequency...

We propose to bring our novel rich media interface called MediaDiver demonstrating new interaction techniques for viewing and annotating multiple view video. The demonstration allows attendees experience moving target selection methods (called Hold Chase), multi-view techniques, automated quality of analysis switch viewpoints follow targets, integrated annotation or authoring meta-content advanced context sensitive transport timeline functions. As users have become increasingly sophisticated...

Computational color constancy (CCC) consists of computing estimates the ambient scene illumination and using them to remove unwanted chromatic distortions. Much research has focused on illuminant estimation for CCC single images, with few attempts leveraging temporal information intrinsic in sequences correlated images (e.g., frames a video), task known as (TCC). The state-of-the-art TCC is network (TCCNet), deep-learning architecture that uses convolutional long short-term memory...

Human-like behavior is crucial for intelligent service robots that are to perform versatile tasks in day life. In this paper, an integrated approach human-like manipulation presented, which addresses realtime three-dimensional (3-D) workspace modeling and accessibility analysis motion planning. The 3-D uses three main principles: identification of global geometric features, substitution recognized known objects by corresponding solid models the database multi-resolution representation...

Afuzzy rule-based system, widely used in the areas of control and pattern recognition, is mostly structured into inference mechanism fuzzy rules expressed terms sets. In this paper, we provide a general framework for inference,

We propose a model based tracking algorithm which can extract trajectory information of target object by detecting and moving from sequence images. The constructs the detected match with successive image frames to track object. use an active characterizes regional structural features such as shape, texture, color, edge. Our adapt itself dynamically so that it non-rigid Such adaptation is made under framework energy minimization. design function embody attributes well its spectral attributes....

This paper describes a real-time tracking system which detects an object entering into the field of view camera and executes detected by controlling servo device so that target always lies at center image frame. In order to detect track moving object, we basically apply model matching strategy. We allow vary dynamically during process it can assimilate variations shape intensities object. also utilize Kalman filter history be encoded state parameters filter. The estimated is then used reduce...

Research into computer vision techniques has far out-paced the development of interfaces (such as APIs) to support techniques' accessibility, especially developers who are not experts in field. We present a new interface, specifically for segmentation methods, designed be application-deveLoper-friendly while retaining sufficient power and flexibility solve wide variety problems. The interface presents at higher level (above algorithms) uses task-based description derived from definitions...

The opacity of deep learning models constrains their debugging and improvement. Augmenting with saliency-based strategies, such as attention, has been claimed to help get a better understanding the decision-making process black-box models. However, some recent works challenged saliency's faithfulness in field Natural Language Processing (NLP), questioning attention weights' adherence true model. We add this discussion by evaluating in-model saliency applied video processing task for first...

We propose a model based tracking algorithm which can extract trajectory information of target object by detecting and moving from sequence images. use an active characterizes regional structural features such as shape, texture, color, edgeness. Our adapt itself dynamically to image so that it track non-rigid object. applied Kalman filter predict motion information. The predicted the was used very efficiently reduce search space in matching process.

Tracking moving objects is one of the most important techniques in motion analysis and understanding, it has many difficult problems to solve. Estimating identifying objects, when background vary dynamically, are especially difficult. It possible under such a complex environment that targets might disappear totally or partially due occlusion by other objects. The Kalman filter been used estimate information use predicting appearance succeeding frames. In this article, we propose another...