In Thailand, the site-specific nutrient management technology, known as “Tailor-made Fertilizer Technology (TFT)”, for rice, maize and sugarcane in the Northeastern region was developed between 1997-2007, using the ...In Thailand, the site-specific nutrient management technology, known as “Tailor-made Fertilizer Technology (TFT)”, for rice, maize and sugarcane in the Northeastern region was developed between 1997-2007, using the concepts of precision agriculture together with an approach of building capacity of small farmers. TFT, also called Smart-farming, comprises four components, namely 1) soil series identification, 2) N-P-K testing by soil test kit, 3) fertilizer recommendations using decision-aids and a simplified version of a complex model and 4) farmer empowerment. The benefit of TFT at the rice field of the Huay Kamin chairman farmer group was one example, the technology has been disseminated to the 80 members with a total planting area of about 320 ha. The results revealed chemical fertilizer reduction of 69%, and rice yield increased some 10% - 20% with the improved fertilizer application method. The farmers were encouraged to establish “Soil Clinics” in their communities. In a Soil Clinic, designated and trained farmer leaders analyze soil samples for member farmers and provide TFT recommendations while providing access to fertilizer materials available for sale at competitive prices. At present, there are about 70 soil clinics in 20 provinces with the support of many government and private sectors.展开更多
为深入了解基于视觉智能感知的畜禽智慧养殖管理与疫病诊断的研究现状,本研究以“深度学习”、“个体检测”、“畜禽身份识别”、“体尺体重评估”、“体温检测”、“行为识别”、“疫病诊断”等为关键词,在Web of Science核心集合、Sci...为深入了解基于视觉智能感知的畜禽智慧养殖管理与疫病诊断的研究现状,本研究以“深度学习”、“个体检测”、“畜禽身份识别”、“体尺体重评估”、“体温检测”、“行为识别”、“疫病诊断”等为关键词,在Web of Science核心集合、Science Direct、CNKI等数据库就1990—2022年已发表的文献进行检索,从5个方面对研究畜禽智慧养殖管理与疫病诊断的方法与技术进行总结、归纳、分析。结果表明:1)畜禽身份识别主要通过畜禽面部识别实现,针对单帧的畜禽面部数据设计无约束方法是未来研究方向。2)畜禽体尺体重智能评估研究中,基于三维点云的畜禽体尺体重高精度快速测量技术是研究的重点。3)由于畜禽疫病数据集的稀缺,基于小样本的畜禽疫病识别技术是突破疫病诊断的关键。4)畜禽体温检测关键是在复杂养殖环境下畜禽热窗的准确定位,通过检测分割算法对热红外模式下的图像进行精准检测。5)日常行为识别主要难点为长时间畜禽密集目标检测与跟踪,并计算其行为轨迹与特点;异常行为通过连续帧间的上下文关系进行识别,主要难点为畜禽异常行为数据稀少性和正负样本不均衡的问题。本综述对基于视觉智能感知的畜禽体温检测、体尺体重评估、行为识别与疫病诊断技术方法进行了研究现状阐述、难点分析和未来趋势展望,为视觉感知技术在畜禽养殖的技术演进和应用发展提供了参考方向。展开更多
文摘In Thailand, the site-specific nutrient management technology, known as “Tailor-made Fertilizer Technology (TFT)”, for rice, maize and sugarcane in the Northeastern region was developed between 1997-2007, using the concepts of precision agriculture together with an approach of building capacity of small farmers. TFT, also called Smart-farming, comprises four components, namely 1) soil series identification, 2) N-P-K testing by soil test kit, 3) fertilizer recommendations using decision-aids and a simplified version of a complex model and 4) farmer empowerment. The benefit of TFT at the rice field of the Huay Kamin chairman farmer group was one example, the technology has been disseminated to the 80 members with a total planting area of about 320 ha. The results revealed chemical fertilizer reduction of 69%, and rice yield increased some 10% - 20% with the improved fertilizer application method. The farmers were encouraged to establish “Soil Clinics” in their communities. In a Soil Clinic, designated and trained farmer leaders analyze soil samples for member farmers and provide TFT recommendations while providing access to fertilizer materials available for sale at competitive prices. At present, there are about 70 soil clinics in 20 provinces with the support of many government and private sectors.
文摘为深入了解基于视觉智能感知的畜禽智慧养殖管理与疫病诊断的研究现状,本研究以“深度学习”、“个体检测”、“畜禽身份识别”、“体尺体重评估”、“体温检测”、“行为识别”、“疫病诊断”等为关键词,在Web of Science核心集合、Science Direct、CNKI等数据库就1990—2022年已发表的文献进行检索,从5个方面对研究畜禽智慧养殖管理与疫病诊断的方法与技术进行总结、归纳、分析。结果表明:1)畜禽身份识别主要通过畜禽面部识别实现,针对单帧的畜禽面部数据设计无约束方法是未来研究方向。2)畜禽体尺体重智能评估研究中,基于三维点云的畜禽体尺体重高精度快速测量技术是研究的重点。3)由于畜禽疫病数据集的稀缺,基于小样本的畜禽疫病识别技术是突破疫病诊断的关键。4)畜禽体温检测关键是在复杂养殖环境下畜禽热窗的准确定位,通过检测分割算法对热红外模式下的图像进行精准检测。5)日常行为识别主要难点为长时间畜禽密集目标检测与跟踪,并计算其行为轨迹与特点;异常行为通过连续帧间的上下文关系进行识别,主要难点为畜禽异常行为数据稀少性和正负样本不均衡的问题。本综述对基于视觉智能感知的畜禽体温检测、体尺体重评估、行为识别与疫病诊断技术方法进行了研究现状阐述、难点分析和未来趋势展望,为视觉感知技术在畜禽养殖的技术演进和应用发展提供了参考方向。
