Water temperature, pH, dissolved oxygen (DO), electrical conductivity (EC), and salinity levels are the critical cultivation factors for freshwater aquaculture. This paper proposes a novel wireless multi-sensor system by integrating the temperature, pH, DO, and EC sensors with an ESP 32 Wi-Fi module for monitoring the water quality of freshwater aquaculture, which acquires the sensing data and salinity information directly derived from the EC level. The information of water temperature, pH, DO, EC, and salinity levels was displayed in the ThingSpeak IoT platform and was visualized in a user-friendly manner by ThingView APP. Firstly, these sensors were integrated with an ESP32 Wi-Fi platform. The observations of sensors and the estimated salinity from the EC level were then transmitted by a Wi-Fi network to an on-site Wi-Fi access point (AP). The acquired information was further transmitted to the ThingSpeak IoT and displayed in the form of a web-based monitoring system which can be directly visualized by online browsing or the ThingView APP. Through the complete processes of pre-calibration, in situ measurement, and post-calibration, the results illustrate that the proposed wireless multi-sensor IoT system has sufficient accuracy, reliable confidence, and a good tolerance for monitoring the water quality of freshwater aquaculture.

In this paper, an efficient instance segmentation scheme based on deep convolutional neural networks is proposed to deal with unconstrained psoriasis images for computer-aided diagnosis. To achieve instance segmentation, the You Only Look At CoefficienTs (YOLACT) network composed of backbone, feature pyramid network (FPN), Protonet, and prediction head is used to deal with psoriasis images. The backbone network is used to extract feature maps from an image, and FPN is designed to generate multiscale feature maps for effectively classifying and localizing objects with multiple sizes. The prediction head is used to predict the classification information, bounding box information, and mask coefficients of objects. Some prototypes generated by Protonet are combined with mask coefficients to estimate the pixel-level shapes for objects. To achieve instance segmentation for unconstrained psoriasis images, YOLACT++ with a pretrained model is retrained via transfer learning. To evaluate the performance of the proposed scheme, unconstrained psoriasis images with different severity levels are collected for testing. As for subjective testing, the psoriasis regions and normal skin areas can be located and classified well. The four performance indices of the proposed scheme were higher than 93% after cross validation. About object localization, the Mean Average Precision (mAP) rates of the proposed scheme were at least 85.9% after cross validation. As for efficiency, the frames per second (FPS) rate of the proposed scheme reached up to 15. In addition, the F1_score and the execution speed of the proposed scheme were higher than those of the Mask Region-Based Convolutional Neural Networks (R-CNN)-based method. These results show that the proposed scheme based on YOLACT++ can not only detect psoriasis regions but also distinguish psoriasis pixels from background and normal skin pixels well. Furthermore, the proposed instance segmentation scheme outperforms the Mask R-CNN-based method for unconstrained psoriasis images.

In this paper, a keyword detection scheme is proposed based on deep convolutional neural networks for personal information protection in document images. The proposed scheme is composed of key character detection and lexicon analysis. The first part is the key character detection developed based on RetinaNet and transfer learning. To find the key characters, RetinaNet, which is composed of convolutional layers featuring a pyramid network and two subnets, is exploited to detect key characters within the region of interest in a document image. After the key character detection, the second part is a lexicon analysis, which analyzes and combines several key characters to find the keywords. To train the model of RetinaNet, synthetic image generation and data augmentation are exploited to yield a large image dataset. To evaluate the proposed scheme, many document images are selected for testing, and two performance measurements, IoU (Intersection Over Union) and mAP (Mean Average Precision), are used in this paper. Experimental results show that the mAP rates of the proposed scheme are 85.1% and 85.84% for key character detection and keyword detection, respectively. Furthermore, the proposed scheme is superior to Tesseract OCR (Optical Character Recognition) software for detecting the key characters in document images. The experimental results demonstrate that the proposed method can effectively localize and recognize these keywords within noisy document images with Mandarin Chinese words.

In this paper, mobile devices were used to estimate the received signal strength indicator (RSSI) of wireless channels with three wireless access points (APs). Using the RSSI, the path loss exponent (PLE) was adapted to calculate the estimated distance among the test points (TPs) and the APs, through the root mean square error (RMSE). Moreover, in this paper, the proposed adaptive PLE (APLE) of the TPs was obtained by minimizing the positioning errors of the PLEs. The training samples of RSSI were measured by TPs for 6 days, and different surge processing methods were used to obtain APLE and to improve the positioning accuracy. The surge signals of RSSI were reduced by the cumulated distribution function (CDF), hybrid Kalman filter (KF), and threshold filtering methods, integrating training samples and APLE. The experimental results show that with the proposed APLE, the position accuracy can be improved by 50% compared to the free space model for six TPs. Finally, dynamic real-time indoor positioning was performed and measured for the performance evaluation of the proposed APLE models. The experimental results show that, the minimum dynamic real-time positioning error can be improved to 0.88 m in a straight-line case with the hybrid method. Moreover, the average positioning error of dynamic real-time indoor positioning can be reduced to 1.15 m using the four methods with the proposed APLE.

This paper originally proposes a wireless multisensor module with illuminance, temperature, relative humidity (RH) and carbon dioxide (CO₂) sensors in an aseptic jar incubator for a solid-state fermentation (SSF) of Cordyceps militaris culture. The light intensity, ambient temperature, RH and CO₂ are the critical cultivation factors of C. militaris. First, these sensors are integrated in a multisensor platform which is installed inside a lid and covered with a high-efficiency particulate air (HEPA) membrane of class H14 for sterilization of bacteria and viruses. The observations of sensors are then transmitted by a wireless XBee network where the slave sensor node is fixed at the top of jar lid and the master radio node receives data and uploads to an on-site monitoring node. The acquired information is further transmitted to an iCloud database and displayed in a web-based monitoring system. The results illustrate the proposed wireless multisensor module was validated with sufficient accuracy, reliable confidence and well-tolerance for C. militaris cultivation biotechnology under aseptic conditions.

In this paper, we investigate the near-far performance of an adaptive full parallel interference cancellation (FPIC) multiuser detection scheme for direct-sequence code-division multiple-access (DS-CDMA) communication systems. Simulation results show that the proposed adaptive FPIC detector is superior to the NLMS-based partial PIC (NLMS-PPIC) scheme with high near-far resistant by effective cancellations of the pernicious multiple access interference (MAI) over a near-far environment