創新智慧影像辨識系統自動化判別蔬果生理指標

  • 刊登日期: 2020-04-06
申請系所(單位) 土壤環境科學系
計畫主持人 林耀東 教授 ( 個人網頁 )
計畫名稱(中文) 創新智慧影像辨識系統自動化判別蔬果生理指標
計畫名稱(英文) Innovative Intelligent Image-recognition System for Automatic Judgement of Biological Indicators of Fruits and Vegetables
共同主持人 吳俊霖 教授 / 資訊科學與工程學系
協同主持人 1. 林慧玲 教授 / 園藝學系 2. 韓斌 教授 / 精密工程研究所
中文摘要 台灣蔬果內銷量佔總產量96%,年產值達1,638億台幣,若將外銷比例提升至50%,蔬果年產值可達8,000億台幣,提升近3至10倍產值,並去化蔬果盛產期價跌窘境。而水果外銷比例低之主因為田間採收蔬果成熟度無法均一,導致運送過程耗損造成腐壞率過高。本計畫將建立獨特創新智慧化蔬果顏色/熟度辨識系統,以傳統農業Lab大數據為基礎,結合先進光學技術與AI人工智慧深度學習功能,自動化擷取蔬果表皮數位顏色影像,替換傳統蔬果顏色辨識系統,並以人工智慧深度學習系統鏈結蔬果生理特性,以即時精準判別蔬果熟化程度及生理指標,大幅提升蔬果產值。
英文摘要 Taiwan′s domestic sales of vegetables and fruits account for 96% of the total output, with an annual output value of NT $ 163.8 billion. If the export ratio is increased to 50%, the annual output value of vegetables and fruits can reach NT $ 800 billion, which will increase the output value by nearly 3 to 10 times. Dilemma. The main reason for the low proportion of fruit exports is that the maturity of harvested vegetables and fruits in the field can not be uniform, which causes the consumption process to cause excessive corruption. This plan will establish a unique and innovative intelligent fruit and vegetable color / maturity identification system, based on traditional agricultural Lab big data, combining advanced optical technology and AI artificial intelligence deep learning functions, to automatically capture digital color images of vegetable and fruit skins, replacing traditional vegetables and fruits The color recognition system and the artificial intelligence deep learning system link the physiological characteristics of vegetables and fruits, so as to accurately determine the ripening degree and physiological indicators of vegetables and fruits in real time, and greatly increase the output value of vegetables and fruits.