Application of Deep Learning to Community-Science-Based Mosquito Monitoring and Detection of Novel Species

Mosquito-borne diseases account for human morbidity and mortality worldwide, caused by the parasites (e.g., malaria) or viruses (e.g., dengue, Zika) transmitted through bites of infected female mosquitoes. Globally, billions of people are at risk of infection, imposing significant economic and public health burdens. As such, efficient methods to monitor mosquito populations and prevent the spread of these diseases are at a premium. One proposed technique is to apply acoustic monitoring to the challenge of identifying wingbeats of individual mosquitoes. Although researchers have successfully used wingbeats to survey mosquito populations, implementation of these techniques in areas most affected by mosquito-borne diseases remains challenging. Here, methods utilizing easily accessible equipment and encouraging community-scientist participation are more likely to provide sufficient monitoring. We present a practical, community-science-based method of monitoring mosquito populations using smartphones. We applied deep-learning algorithms (TensorFlow Inception v3) to spectrogram images generated from smartphone recordings associated with six mosquito species to develop a multiclass mosquito identification system, and flag potential invasive vectors not present in our sound reference library. Though TensorFlow did not flag potential invasive species with high accuracy, it was able to identify species present in the reference library at an 85% correct identification rate, an identification rate markedly higher than similar studies employing expensive recording devices. Given that we used smartphone recordings with limited sample sizes, these results are promising. With further optimization, we propose this novel technique as a way to accurately and efficiently monitor mosquito populations in areas where doing so is most critical.

Using a blind test to assess the discriminant power of morphological traits to distinguish between similar shrew species

The Canary Islands are home to a large variety of endemic fauna. The Canary shrew (Crocidura canariensis) has a distribution restricted to Fuerteventura, Lanzarote and the islets of Lobos and Montaña Clara. One of the main threats to the insular fauna is the arrival of exotic species. The greater white-toothed shrew (Crocidura russula) is an easily transportable animal and a potential competitor for C. canariensis. Therefore, C. russula should be taken into account in the management protocols for invasive species. One of the most easily applicable techniques for detecting shrews is the analysis of pellets. This study aims to assess which morphological characters are diagnostic and easy to use when identifying both species of shrews. For this purpose, a blind specific assignment has been made using seven previously described characters and another three added in the present study. The results show that the observer’s experience did not improve the correct identification rate and that only three of the evaluated characters have a high discriminant capacity. Finally, it was found that the combined use of the maximum number of characters and the identification by two independent observers reduces the probability of making a mistake in the determination to minimum values.

Determination of Geographical Origin of Wuchang Rice with the Geographical Indicator by Multielement Analysis

The study aims to investigate whether the multielement analysis result can be used as a fingerprint to identify the geographical origin of Wuchang rice. The element contents of rice and soil samples from three regions in China (Wuchang, Qiqihar, and Jiamusi) were analyzed. The concentrations of 16 elements (Na, Mg, Al, K, Ca, V, Mn, Fe, Co, Cu, Zn, As, Rb, Sr, Cd, and Pb) in 194 rice samples and 112 soil samples from the harvest season in 2013 and 2014 were determined. The analysis of variance and linear discriminant analysis were performed to analyze the variation among regions and rice genotypes and classify the geographical origins of rice. Only the element of Cu showed significant differences among different genotypes. In the discriminant analysis, the overall correct identification rates of the rice samples obtained in 2013 and 2014 were, respectively, 96.6% and 89.6% and the overall correct identification rate for Wuchang rice reached 100%.

Fusion-Based Hand Geometry Recognition Using Dempster–Shafer Theory

This paper presents a new technique for user identification and recognition based on the fusion of hand geometric features of both hands without any pose restrictions. All the features are extracted from normalized left and right hand images. Fusion is applied at feature and also at decision level. Two probability-based algorithms are proposed for classification. The first algorithm computes the maximum probability for nearest three neighbors. The second algorithm determines the maximum probability of the number of matched features with respect to a thresholding on distances. Based on these two highest probabilities initial decisions are made. The final decision is considered according to the highest probability as calculated by the Dempster–Shafer theory of evidence. Depending on the various combinations of the initial decisions, three schemes are experimented with 201 subjects for identification and verification. The correct identification rate is found to be 99.5%, and the false acceptance rate (FAR) of 0.625% has been found during verification.

The Italian External Quality Assessment (EQA) program on urinary sediment: results of the period 2012–2015

Manual microscopy still represents the gold standard for urinary sediment (US) examination. We report the results obtained in the period 2012–2015 by the EQA Italian program on US, which today involves about 260 laboratories.The program includes four surveys per year. In two surveys, participants are asked to supply identification and clinical association of US particles. In two other surveys, they are asked to supply the diagnosis of clinical cases, presented with images, some key laboratory findings and a short clinical history. Sixty-six images of US particles (21 cells, 2 lipids, 21 casts, 10 crystals, 3 microorganisms, 15 contaminants) and seven clinical cases were presented.The correct identification rate for each category of particles, in decreasing order, was: micro-organisms (mean±SD: 92.4%±4.5%), lipids (92.0%±1.8%), casts (82.8%±8.8%), crystals (79.4%±29.8%), cells (77.3%±13.5%), and contaminants (70.9%±22.2%). For 13 particles, a correct clinical association was indicated by 91.5%±11.7% of participants, while it was 52.7% for particles associated with urinary tract infection. For clinical cases, due to a high rate of particles misidentification, only 44.3%±10.1% of participants achieved access to clinical diagnosis, which was then correctly indicated by 92.5%±5.3% of them.The results of the EQA program confirm that, while some US particles are well known in terms of identification, clinical association and clinical meaning, others particles still are not, and this represents an important reason to encourage EQA programs on US.

An experimental study investigating the ability of volunteers to identify squirrel species from tail–hair samples

Hair–tubes, collecting nape hairs, are widely used for establishing the presence of red (Sciurus vulgaris) and grey (Sciurus carolinensis) squirrels. However it is time–consuming and prone to identification errors. An alternative is to collect tail hairs from sticky pads on baited poles. However, there is no evidence concerning identification accuracy of tail hairs. This study reports an experiment in which subjects underwent a short training session before identifying hair samples from four species. There was a 96.5% correct identification rate for grey squirrel hairs, and 77.5% for red squirrels, which suggests that tail hairs collection may provide a quick, easy and accurate method of identification for both species.

The Algorithm of Digital Modulation Recognition Based on Stochastic

The traditional digital modulation recognition methods have high requirements for the SNR. This paper presented a good identification method. The quadratic stochastic resonance for signal processing cascade was proposed, then the signal characteristics of high order cumulants were extracted and the support vector machine was used for classifier, simulation results showed that the correct identification rate was up to 88% or more when the SNR was low as-8dB. The bistable stochastic resonance treatment can effectively improve the SNR.

The utility of the ENTERORapid 24 kit for the identification of P. multocida and M. haemolytica

ENTERORapid 24 kit (PLIVA-Lachema, Czech Republic) was used for the identification of 321 strains isolated from the respiratory tract of different animal species in the CzechRepublic and Ethiopia. A total of 207 ( 64.5%) strains were identified at the species level within 4 to 8 hours of incubation. In the same way, 39 (12.1%) strains were successfully classified at the genus level. The remaining strains (23.4%) were not identified nor classified to the family Pasteurellaceae. On the other hand, the accuracy of the ENTERORapid 24 kit for the identification of P. multocida and M. haemoyltica was observed using 9 reference strains and the identification results were compaired with the results of the RapiD 20E kit (bioMérieux, France), which required an overall examination time of 4 hours. According to our observation, the ENTERO Rapid 24 kit is the fastest system forthe identification of P. multocida and M. haemolytica strains within 4 to 8 hours with a correct identification rate at thespecies level, with and without additional tests. For these reasons, we propose its modification for rapid identificaton of P. multocida, M. haemolytica and related bacterial species from the family Pasteurellaceae isolated from different animal species.

Fuel Identification Based on the Least Squares Support Vector Machines

The identification of the fuel types plays an important role in ensuring the safety and economics of the power plants. In order to obtain the flame signal in the process of combustion, a flame detection system is designed and a laboratorial platform is constructed. This paper extracts the signal parameters—the mean, the peak-peak value, the flicker frequency, and the flicker intensity —and takes them as the characteristic quantities of the flame signal. Based on the least squares support vector machines (LSSVM), an efficient method of identifying the flame types is developed. The result of the identification is more ideal, with the correct identification rate up to 100%. This shows that the method combined the four characteristic quantities with the LSSVM can obtain a good result in the identification of the fuel types.

The Design of a Car License Plate Identification System Based on AdaBoost Algorithm

Design and implement a car license plate identification system with the applications of Viola and Jones algorithm. This algorithm which is based on the AdaBoost method is trained and optimized for the best performance using large database of car license plate images. The final license plate identification system obtained a cascade of classifiers consisting of 8 stages with 1310 Haar-like features. Once the license plates have sufficient visibility and there are no other objects similar to the plate in images, this system operates perfectly and shows high correct identification rate with low false positive rate. And as integral image allows the Haar-like features to be calculated very fast, the system also finished the identification rapidly.