Food recognition using neural network classifier and multiple hypotheses image segmentation

ABSTRACT

This paper proposes the multiple-hypotheses image segmentation and feed-forward neural network classifier for food recognition to improve the performance. Initially, the food or meal image is given as input. Then, the segmentation is applied to identify the regions, where a particular food item is located using salient region detection, multi-scale segmentation, and fast rejection. Then, the features of every food item are extracted by the global feature and local feature extraction. After the features are obtained, the classification is performed for each segmented region using a feed-forward neural network model. Finally, the calorie value is computed with the aid of (i) food volume and (ii) calorie and nutrition measure based on mass value. The experimental results and performance evaluation are validated. The outcome of the proposed method attains 0.947 for Macro Average Accuracy (MAA) and 0.959 for Standard Accuracy (SA), which provides better classification performance.     

Survey of domestic food purchases and related home handling practices in the Abruzzo region (central Italy): Data collection and analysis through a language-independent classification system

A telephone survey was carried out on 13,486 randomly selected households located in the Abruzzo region (central Italy). Three questionnaires were specifically designed in relation to three different groups of foods (Meat and meat products, Fishery products, Fruit and vegetables). Questions were mainly focused on the amount and number of purchases, type of retailer, and home food handling practices with respect to the week prior to the interview. Data were classified according to a multilingual thesaurus system (LanguaL). Results allowed to estimate domestic purchases (in kg) per capita for different food categories. The category “Red meat/Cattle” accounted for a large part of purchases in the “meat and meat products” group (10.9 kg per capita/year; 32.2% of purchases in the group) while the category “Fish or related organism/Fish” was the most purchased in the “fishery products” group (6.9 kg per capita/year; 63.3%). The aggregation of more detailed characteristics enabled the identification of popular categories of foods, such as “Red meat/Cattle/Divided into pieces” (25% of the “meat and meat products” group). Significant differences (p < 0.05) were found between the four different provinces (L'Aquila, Chieti, Pescara and Teramo) with respect to “Poultry/Chicken” and “Red meat/Cattle” categories. Householders were also asked about post-purchase food handling practices that might be hazardous to food safety. More than 20% of those surveyed stated that they thawed frozen meat at room temperature. The degree of doneness after cooking of different food categories was generally high: over 90% of products purchased in the majority of meat and fish categories were properly cooked. However a noticeable proportion of householders (about 15%) reported medium or rare cooking of “Red meat/Cattle” and “Red meat/Swine”. Differences (p < 0.05) were also found between consumers of different ages, with people over 65 years old being more prone to freeze meat and cook it thoroughly.The survey was carried out in a specific geographic area and on a statistically significant sample of households, thus allowing a collection of data on domestic habits relating to food purchases and home food handling practices. This information should be included in the framework of quantitative microbial risk assessment (QMRA) models as a measure of the actual exposure of consumers to pathogenic microorganism. The LanguaL system also proved to be a practical language-independent method useful not only to identify and describe food items but also to classify them according to specific food safety characteristics.     

Performance discrepancies on the California Verbal Learning Test-Second Edition (CVLT-II) in the standardization sample.

The standardization data for the California Verbal Learning Test-Second Edition (CVLT-II; D. C. Delis, J. H. Kramer, E. Kaplan, & B. A. Ober, 2000) were used to evaluate the base rate of 6 specific discrepancies between various key variables. The results indicated that CVLT-II performance discrepancies should equal or exceed 1 or 1.5 z score points (depending on the individual comparison) in the hypothesized direction to be considered potentially unusual. However, because about 1 in 3 persons in the standardization sample displayed at least 1 such large discrepancy, it is concluded that these base rates should be viewed only as a starting point for interpretation. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Using a multicultural lens to understand illnesses among Haitians living in America.

Currently, ethnic and racial minority individuals represent a large proportion of the U.S. population, and researchers expect that they will represent the majority of the population by 2050. This shift in U.S. demographics calls for a greater awareness and integration of cultural issues into the assessment and treatment of ethnically and linguistically diverse clients. This article examines the unique beliefs and manifestations of illnesses among Haitians in connection with the American Psychological Association's (APA, 2002) Multicultural Guidelines. The authors use a common culture-bound syndrome, Séizisman, to illustrate the cultural beliefs, assessment, and treatment of illnesses among Haitians. In so doing, they demonstrate how to incorporate the APA Multicultural Guidelines into treatment with clients of diverse cultural and racial backgrounds. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

横泉水库供水发电洞工程地质分析与评价

通过大量野外工程地质勘察与测试工作，在此基础上给出了横泉水库供水发电洞的工程地质分析与评价，从而使该水库的施工有了充分的理论依据。     

管道风险管理方法研究

按照管道风险管理的流程分别对管道风险评价、风险控制和决策支持、效能测试和响应进行了论述。针对目前国内管道行业的情况，提出了进行管道风险评价的有效方法及维护措施。着重介绍了国外管道风险可接受标准的情况，作为国内制定管道风险评价标准的参考。     

国际石油合同中的风险与对策

石油公司从事跨国石油勘探开发活动，必然受到国际石油合同的影响和制约。国际石油合同包括租让制合同、产品分成合同、风险服务合同、回购合同、联合经营和国家石油公司私有化6类。石油合同风险分为地质风险、技术风险、财务风险、商业风险和政治风险5种类型。通过全面分析国际石油合同及其风险的类型和特点，认为在油气勘探开发项目的不同阶段风险管理的重点不同，石油公司必须建立科学的投资决策程序，采取定量与定性相结合的风险分析和投资决策方法，合理利用资源国财税制度等措施规避和降低风险，对中国的石油公司从事跨国石油勘探开发具有重要意义。     

低勘探程度区域油气资源评价方法

目前中国油气资源勘探和开发面临后备可采储量不足，低勘探程度区域受到重视。为了提高低勘探程度区域的资源评价精度，在前人评价的基础上，通过研究并利用勘探程度较高区域的资源评价结果，求准低勘探程度区域的油气聚集量，进而模拟求出其烃排聚系数、储量密度系数、单储系数和单生系数等关键参数，提出了利用练台地质推断法对低勘探程度区域进行评价，并对该方法评价的风险进行了分析。综合地质推断法的成果更能准确反映一个盆地、坳（凹）陷或区带的油气潜力及其分布特征。     

应用API 581对输气站场进行定量风险评价

目前专门针对输气站场的风险评价方法还不多，现有的手段也只是根据国家法律法规得出定性的评价结果，然后按照评价结果对照得出整改方案。为此，对输气站场的定量风险评价工作提出了一套新方法，得到了定量的和更加客观的风险排序，为管道站场的完整性管理工作提供了有力的支持；并以陕京输气管道榆林压气站为例，应用美国石油学会标准API 581（基于风险检测的基本源文件）对站内设施进行了风险评价，计算了站内设备和管线的失效后果、失效概率，得到了定量风险值，进而分析了风险值产生差异的原因。最后，针对我国输气管道风险评估工作提出了看法和建议。     

天然气产业链下游市场风险评价模型探讨

随着我国天然气主干管网的逐步形成，天然气产业链下游市场正在进入快速成长阶段，建立风险预警系统的重要性开始显现，而风险评价模型是其基础工作之一。文章在天然气产业链下游区域市场风险评价指标体系的基础上，综合运用层次分析法和灰色系统理论，建立了对天然气产业链下游区域市场风险评价的模型，并以各区域市场的用气量占全国总用气量的比重为权重，建立起完整的市场风险评价模型，可为评价天然气产业链下游市场整体市场风险时参考。