Optimization of IoT-Based Motion Intelligence Monitoring System

We design and implement an intelligent IoT-based motion monitoring system to realize the monitoring of three important parameters, namely, the type of movement, the number of movements, and the period of movement in physical activities, and optimize the system to support the simultaneous use by multiple users. Considering the motion monitoring scenario for smart fit, the framework of an IoT-based motion monitoring system is proposed. The framework contains components such as active acquisition nodes, wireless access points, data processing servers, and terminals. In terms of algorithm optimization, research related to active pattern recognition and periodic calculation methods is conducted. For active pattern recognition, two types of classification algorithms with different complexity are proposed based on Support Vector Machine (SVM) and deep neural networks, respectively, to adapt to scenarios with different computational capabilities. For period calculation, a method based on over-zero detection and wavelet transform is proposed to count the number of actions and calculate the period of each action. In 100 times action cycle calculation experiments, the count statistic calculation method achieves 100% calculation accuracy and the active cycle calculation results are close to the real value, which proves the effectiveness of the cycle calculation method. The system provides a multiuser-oriented communication method and realizes accurate and reliable human movement monitoring, which has a wide application prospect in the fields of physical education and rehabilitation training.

An approach to color-based sorting of bulk materials with automated estimation of system parameters

In this paper, we present a flexible method for color-based sorting of bulk materials. It is based on semantically meaningful color features that are constructed from a set of training images. First, estimates of color-occurrence frequencies of different materials are derived from the training images and fused into color classes, which are then used to classify individual pixels. An object descriptor is built as count statistic over the color classes appearing in the object image. This descriptor has many advantages: it is compact and very fast to compute, invariant to scale and rotation, has a very clear, intuitive interpretation, and can be used with simple rule-based classifiers. However, tuning the parameters that govern the feature construction process is laborious and requires a lot of experience on part of the system operator. To overcome this shortcoming, we automatically learn the parameters using genetic algorithms. We apply our method to wine grape sorting problems to show that this approach outperforms a human expert. At the same time, it takes considerably less effort on the human part and frees the expert to attend to other tasks. Furthermore, the system allows non-experts to successfully put a sorting machine in operation.

Spatial and Spatiotemporal Pattern Analysis of Coconut Lethal Yellowing in Mozambique

Coconut lethal yellowing (LY) is caused by a phytoplasma and is a major threat for coconut production throughout its growing area. Incidence of LY was monitored visually on every coconut tree in six fields in Mozambique for 34 months. Disease progress curves were plotted and average monthly disease incidence was estimated. Spatial patterns of disease incidence were analyzed at six assessment times. Aggregation was tested by the coefficient of spatial autocorrelation of the β-binomial distribution of diseased trees in quadrats. The binary power law was used as an assessment of overdispersion across the six fields. Spatial autocorrelation between symptomatic trees was measured by the BB join count statistic based on the number of pairs of diseased trees separated by a specific distance and orientation, and tested using permutation methods. Aggregation of symptomatic trees was detected in every field in both cumulative and new cases. Spatiotemporal patterns were analyzed with two methods. The proximity of symptomatic trees at two assessment times was investigated using the spatiotemporal BB join count statistic based on the number of pairs of trees separated by a specific distance and orientation and exhibiting the first symptoms of LY at the two times. The semivariogram of times of appearance of LY was calculated to characterize how the lag between times of appearance of LY was related to the distance between symptomatic trees. Both statistics were tested using permutation methods. A tendency for new cases to appear in the proximity of previously diseased trees and a spatially structured pattern of times of appearance of LY within clusters of diseased trees were detected, suggesting secondary spread of the disease.