MTSTR: Multi-task learning for low-resolution scene text recognition via dual attention mechanism and its application in logistics industry.

Recognizing texts in images plays an important role in many applications, such as industrial intelligence, robot vision, automatic driving, command assistance, and scene understanding. Although great progress has been achieved in various fields, research on complex systems modeling using text recognition technology requires further attention. To address this, we propose a new end-to-end multi-task learning method, which includes a super-resolution branch (SRB) and a recognition branch. To effectively learn the semantic information of images, we utilize the feature pyramid network (FPN) to fuse high- and low-level semantic information. The feature map generated by FPN is then delivered separately to the super-resolution branch and the recognition branch. We introduce a novel super-resolution branch, the SRB based on the proposed dual attention mechanism (DAM), designed to enhance the capability of learning low-resolution text features. The DAM incorporates the residual channel attention to enhance channel dependencies and the character attention module to focus on context information. For the recognition branch, the feature map generated by FPN is fed into an RNN sequence module, and an attention-based decoder is constructed to predict the results. To address the issue of low-resolution text recognition in numerous Chinese scenes, we propose the Chinese super-resolution datasets instead of relying on traditional down-sampling techniques to generate training datasets. Experiments demonstrate that the proposed method performs robustly on low-resolution text images and achieves competitive results on benchmark datasets.

A self-learning mean optimization filter to improve bluetooth 5.1 AoA indoor positioning accuracy for ship environments.

As COVID-19 is still spreading globally, the narrow ship space makes COVID-19 easier for the virus to infect ship passengers. Tracking close contacts remains an effective way to reduce the risk of virus transmission. Therefore, indoor positioning technology should be developed for ship environments. Today, almost all smart devices are equipped with Bluetooth. The Angle of Arrival (AoA) using Bluetooth 5.1 indoor positioning technology is well suited for ship environments. But the narrow ship space and steel walls make the multipath effect more pronounced in ship environments. This also means that more noises are included in the signal. In the Uniform Rectangular Array (URA) type receiving antenna array, elevation and azimuth angles are two important parameters for the AoA indoor positioning technology. Elevation and azimuth angles are unstable because of the influence of noise. In this paper, a Self-Learning Mean Optimization Filter (SLMOF) is proposed. The goal of SLMOF is to find the optimal elevation and azimuth angles as a way to improve the Bluetooth 5.1 AoA indoor positioning accuracy. The experimental results show that the Root Mean Square Error (RMSE) of SLMOF is 0.44 m, which improves the accuracy by 72% compared to Kalman Filter (KF). This method can be applied to find the optimal average in every dataset.

A close contact identification algorithm using kernel density estimation for the ship passenger health.

COVID-19 has been spread globally, with ships posing a significant challenge for virus containment due to their close-quartered environments. The most effective method for preventing the spread of the virus currently involves tracking and physically isolating close contacts. In this paper, we propose the Close Contact Identification Algorithm (CCIA). The probability density of user location points may be higher in a certain spatial range such as a cabin where there are more location points. The characteristics of CCIA include using Kernel Density Estimation (KDE) to calculate the probability density of each user location point and seeking the maximum Euclidean distance between location points in each cluster for merging clusters. CCIA is capable of calculating the probability density of each location point, a feature that other clustering algorithms, such as Kmeans, Hierarchical, and DBSCAN, cannot achieve. The contribution of CCIA is using the probability density of each location point to identify close contacts in ship environments. The performance of CCIA shows more accurate clustering compared to Kmeans, Hierarchical, and DBSCAN. CCIA can effectively identify close contacts and enhance the capabilities of user devices in mitigating the spread of COVID-19 within ship environments.