OTHER FUNCTIONS SBIR TOPIC 308, PHASE I: THE MOBILE FOOD INTAKE PHOTO STORAGE AN

其他功能 SBIR 主题 308，第一阶段：移动食物摄取照片存储和

• 批准号: 8554263
• 负责人: RICK WEISS
• 金额: $ 20万
• 依托单位: VIOCARE, INC.
• 依托单位国家: 美国
• 财政年份: 2012
• 资助国家: 美国
• 起止时间: 2012-09-28 至 2013-06-27
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/8554263
• 关键词: Car Phone; Clinical Research; Computer Vision Systems; Data; Diet; Dietary intake; Documentation; Eating; Environment; Evaluation; Food; Genes; Health; Image; Image Analysis; Measurement; Measures; Phase; Reporting; Small Business Innovation Research Grant; System; Three-Dimensional Image; Time; Visual; Voice; feeding; innovative technologies; novel; prototype; speech recognition; tool

This proposal describes plans to enhance Viocare¿s Mobile Food Intake Visual and Voice Recognizer (FIVR) System. FIVR, an active Genes, Environment and Health Initiative (GEI) project, is a novel combination of innovative technologies including computer vision and speech recognition to measure dietary intake using a mobile phone. Version 1 of FIVR uses a mobile phone¿s embedded camera to capture a short video of food to be consumed. The food to be eaten is annotated verbally on the mobile phone by the user. These video and audio files are sent to a backend server for real-time food recognition and portion size measurement through speech recognition and image analysis. This project will develop specifications to extend FIVR¿s capabilities to standardize, store, and analyze more diverse food images, such as 3D photos; to collect other food data; to enhance the analysis tools; and for interfaces to a variety of clinical/research systems. The FIVR Version 2 functional prototype will be developed to use 3D dietary images as input. A final evaluation of the FIVR V2 prototype will be conducted to assess the accuracy and feasibility of the 3D image diet capture with a group of 9 subjects in a controlled feeding study.

该提案描述了增强Viocare移动的食物摄入视觉和语音识别（FIVR）系统的计划。FIVR是一个活跃的基因、环境和健康倡议（GEI）项目，是一种创新技术的新组合，包括计算机视觉和语音识别，使用移动的手机测量饮食摄入量。FIVR的第一个版本使用移动的手机的嵌入式摄像头来捕捉食物的短视频。用户在移动的电话上口头注释要吃的食物。这些视频和音频文件被发送到后端服务器，用于通过语音识别和图像分析进行实时食物识别和份量测量。该项目将制定规范，以扩展FIVR的能力，以标准化，存储和分析更多样化的食物图像，如3D照片;收集其他食物数据;增强分析工具;并用于各种临床/研究系统的接口。将开发FIVR第2版功能原型，以使用3D饮食图像作为输入。将对FIVR V2原型进行最终评价，以评估受控喂养研究中9名受试者的3D图像饮食捕获的准确性和可行性。