Hashing technologies have been widely applied for large-scale multimodal retrieval tasks owing to their excellent performance in search and storage tasks. Although some effective hashing methods proposed, it is still difficult handle the intrinsic linkages that exist among different heterogeneous modalities. Moreover, optimizing discrete constraint problem through a relaxation-based strategy results large quantization error leads suboptimal solution. In this article, we present novel...

Hashing has been widely applied in large-scale multimodal retrieval by mapping heterogeneous modalities data into binary codes. However, most cross-modal hashing methods cannot make the of semantic information to construct association relations sample pairs, resulting unsatisfactory accuracy. Concept lattice is a powerful tool for mining and retrieval, all we know, this first time combine formal concept analysis hash learning improve performance. In paper, propose novel framework Asymmetric...

Unsupervised deep cross-modal hash retrieval aims to map multi-modal features into binary codes without labels, which is of interest due its storage efficiency, query speed and convenient applications. However, existing approaches suffer from two main limitations: (1) Slightly insufficient consideration text instance similarity, along with independent or redundant fusion learn similarity information. (2) They ignore the noisy adjacent correlations between instances, leading a lack...

Deep convolutional neural networks (DCNNs) are currently popular in human activity recognition (HAR) applications. However, the face of modern artificial intelligence sensor-based games, many research achievements cannot be practically applied on portable devices (i.e., smart phone, VR/AR). DCNNs typically resource-intensive and too large to deployed devices, thus, this limits practical application complex detection. In addition, since do not possess high-performance graphic processing...

Hashing techniques have recently been successfully applied to solve similarity search problems in the information retrieval field because of their significantly reduced storage and high-speed capabilities. However, hash codes learned from most recent cross-modal hashing methods lack ability comprehensively preserve adequate information, resulting a less than desirable performance. To this limitation, we propose novel method termed Nonlinear Robust Discrete (NRDH), for retrieval. The main...

Mushroom picking robot could significantly decrease the labor cost during production process. As well as humans, robots need to know what mushroom is, where it and finally harvest it. This paper proposes an accurate real-time capable object detection localization approach for use on oyster robot. Detection information of a neural network is combined with depth from RGB-D camera, which mounted platform. The SSD algorithm used convolutional network. In order find detected in 3D environment,...

With the rapid growth of multimedia data (e.g., image, audio, and video) on Web, learning-based hashing techniques, such as deep supervised hashing, have proven to be very efficient for large-scale search. The recent successes seen in methods are largely due success methods. However, there some limitations previous learned hash codes containing repetitive highly correlated information). In this paper, we propose a novel method, named attention-guided (DAgH). DAgH is implemented using two...

The ripening information of tomato fruit in greenhouse environment is closely related to production operation. Currently, mainly carried out by manual inspection. In this paper, we present a new real-time method detection and its maturity measurement natural greenhouse. More than 25,000 fruits were tested 1005 images Yolov4 the identification accuracy was 95%. After construction single data set, color proportion analysis RGB features analyzed. Combined with normalization noise reduction...

Human activity recognition (HAR) using deep neural networks has become a hot topic in human–computer interaction. Machines can effectively identify human naturalistic activities by learning from large collection of sensor data. Activity is not only an interesting research problem but also many real-world practical applications. Based on the success residual achieving high level aesthetic representation automatic learning, we propose novel asymmetric network, named ARN. ARN implemented two...

The contact window scheduling technology supports various space mission requirements of the satellite information network. continuous emergence new missions challenges scarce scheduling. In this article, We propose a dynamically updated microcloud structure based on digital twin and multiagent system technology. Then, we build an on-satellite autonomous model design efficient solution algorithm to solve maximum revenue problem inefficient task caused by dynamic change priority position over...