Research on clouds has an enormous influence on sky sciences and related applications, and cloud classification plays an essential role in it. Much research has been conducted which includes both traditional machine learning approaches and deep learning approaches. Compared with traditional machine learning approaches, deep learning approaches achieved better results. However, most deep learning models need large data to train due to the large number of parameters. Therefore, they cannot get high accuracy in case of small datasets. In this paper, we propose a complete solution for high accuracy of classification of cloud image patches on small datasets. Firstly, we designed a suitable convolutional neural network (CNN) model for small datasets. Secondly, we applied regularization techniques to increase generalization and avoid overfitting of the model. Finally, we introduce a model average ensemble to reduce the variance of prediction and increase the classification accuracy. We experiment the proposed solution on the Singapore whole-sky imaging categories (SWIMCAT) dataset, which demonstrates perfect classification accuracy for most classes and confirms the robustness of the proposed model.

This paper presents a sound source distance estimation (SSDE) method using a convolutional recurrent neural network (CRNN). We approach the sound source distance estimation task as an image classification problem, and we aim to classify a given audio signal into one of three predefined distance classes—one meter, two meters, and three meters—irrespective of its orientation angle. For the purpose of training, we create a dataset by recording audio signals at the three different distances and three angles in different rooms. The CRNN is trained using time-frequency representations of the audio signals. Specifically, we transform the audio signals into log-scaled mel spectrograms, allowing the convolutional layers to extract the appropriate features required for the classification. When trained and tested with combined datasets from all rooms, the proposed model exhibits high classification accuracies; however, training and testing the model in separate rooms results in lower accuracies, indicating that further study is required to improve the method’s generalization ability. Our experimental results demonstrate that it is possible to estimate sound source distances in known environments by classification using the log-scaled mel spectrogram.

Abstract:In model-based tracking, an accurate 3D model of a target object or scene is mostly assumed to be known or given in advance, but the accuracy of the model should be guaranteed for accurate pose estimation. In many application domains, on the other hand, end users are not highly distracted by tracking errors from certain levels of modeling errors. In this paper, we examine perceptual tracking errors, which are predominantly caused by modeling errors, on subjective evaluation and compare them to computational tracking errors. We also discuss the tolerance of modeling errors by analyzing their permissible ranges.

No abstract

No abstract

No abstract

This paper presents a novel hand-based interface that provides natural interactions with virtual objects augmented on a user's palm. In the interface, virtual objects are easily selected by drawing predefined patterns on a palm of one hand with a finger of the other hand and recognizing the patterns. The selected virtual objects are augmented on the palm by estimating palm poses. To efficiently detect the patterns drawn with the finger, its trajectories are tracked using a wearable type of a visual marker. By estimating a wide range of palm poses and recognizing finger gestures, a user can naturally interact with the virtual objects augmented on the palm at various viewpoints. The feasibility of the interface is verified through our experimental results.

This paper proposes a method for 3-D sound source localization (SSL) using region selection and TDOA. 3-D SSL involves the estimation of an azimuth angle and an elevation angle. With the aim of reducing the computation time, we compare signal energies to select one out of three regions. In the selected region, we compute only one TDOA value for the azimuth angle estimation. Also, to estimate the vertical angle, we choose the higher energy signal from the selected region and pair it up with the elevated microphone's signal for TDOA computation and elevation angle estimation. Our experimental results show that the proposed method achieves average error values of 0.778° in azimuth and 1.296° in elevation, which is similar to other methods. The method uses one energy comparison and two TDOA computations therefore, the total processing time is reduced.키워드 : 방위각, 높이, 신호 에너지, TDOA