環保專案成果報告查詢系統

專案計畫編號
經費年度	113
計畫經費	5,200 千元
專案開始日期	2024/03/15
專案結束日期	2024/12/31
專案主持人	周傳鈴
主辦單位	國環院檢測技術中心
承辦人	張世忠
執行單位	藻田生態有限公司
專案分類	科技計畫、環境監測及檢測
中文關鍵字	浮游藻類、生物指標、人工智慧技術
英文關鍵字	Phytoplankton, Biological indexes, Artificial intelligence technique
協同主持人
共同主持人
計畫聯絡人	周傳鈴
計畫聯絡電話	0910738552
計畫聯絡信箱	searchalgae@gmail.com

中文摘要	水資源是全球最重要的自然資源，水資源的維護及建立為保障社會和經濟發展的關鍵，為避免水源水質破壞因應之水源保護政策致訂有嚴格之規範，並且有系統制定目標進行管理。淨零政策之環境風險評估生物指標建立-植物性浮游生物及藻類AI辨識系統計畫，即為維護良好水域，除了建立最新台灣水域常見的藻類圖冊，並結合新興科技AI學習、創造和影像辨識軟件，以AI辨識系統新趨勢，推動藻類AI辨識系統來進行水質監測。計畫有兩個具體目標，目標一、建立臺灣植物性浮游生物圖片圖冊，本期期末階段已完成120個物種，包括摘述物種特徵、常出現地點以及水環境條件等相關資訊。目標二、製作植物性浮游生物人工智慧(AI)辨識軟體，期中階段已收集建置50種(屬)藻類數位影像，包括確定之學名或屬名或科名，接物鏡使用倍數，達到初步的訓練條件。建立之臺灣植物性浮游生物圖冊，預期能給國內有關藻類鑑識工作人員提供最新藻類圖鑑科普教育之推廣外，配合人工智慧AI辨識系統，將可以大幅降低藻類鑑識之困難度與時間成本，進而大幅提升藻種判定之精確度，對於未來緊急突發性水汙染事件，可利用此創新的藻類鑑識技術，發揮水質預警功能。
中文關鍵字	浮游藻類、生物指標、人工智慧技術

中文摘要

水資源是全球最重要的自然資源，水資源的維護及建立為保障社會和經濟發展的關鍵，為避免水源水質破壞因應之水源保護政策致訂有嚴格之規範，並且有系統制定目標進行管理。淨零政策之環境風險評估生物指標建立-植物性浮游生物及藻類AI辨識系統計畫，即為維護良好水域，除了建立最新台灣水域常見的藻類圖冊，並結合新興科技AI學習、創造和影像辨識軟件，以AI辨識系統新趨勢，推動藻類AI辨識系統來進行水質監測。計畫有兩個具體目標，目標一、建立臺灣植物性浮游生物圖片圖冊，本期期末階段已完成120個物種，包括摘述物種特徵、常出現地點以及水環境條件等相關資訊。目標二、製作植物性浮游生物人工智慧(AI)辨識軟體，期中階段已收集建置50種(屬)藻類數位影像，包括確定之學名或屬名或科名，接物鏡使用倍數，達到初步的訓練條件。建立之臺灣植物性浮游生物圖冊，預期能給國內有關藻類鑑識工作人員提供最新藻類圖鑑科普教育之推廣外，配合人工智慧AI辨識系統，將可以大幅降低藻類鑑識之困難度與時間成本，進而大幅提升藻種判定之精確度，對於未來緊急突發性水汙染事件，可利用此創新的藻類鑑識技術，發揮水質預警功能。

中文關鍵字

浮游藻類、生物指標、人工智慧技術

類型	檔名	檔案大小	說明	最新版公開日期
期末報告	113國環院-AI成果報告(定稿本).pdf	3MB		2025/1/31 上午 12:00:00

摘要	Water is the most important natural resource on the earth. The appropriate exploitation and conservation of water resources are critical to ensure social security and economic development. Therefore, stringent regulations have been promulgated through the water resource protection policies, and systematic goals have been set for management. This project including establishing a pictorial album of common algal species found in Taiwan’s water bodies and algal recognition system based on self-learning and creating AI algorithms, aims at upgrading water quality monitoring technique by using the new AI-assisted recognition technique. The first objective of this project is to establish a pictorial agal album, which includes the pictures and their characteristics of 120 phytoplankton species, and the information about their typical locations, the water environmental conditions and more. The second objective is to establish an AI-assisted recognition system for algal species. As of the mid-project stage, the system has been tested and met the preliminary training performance requirement, including the digital images of 50 algal species (genera) with confirmed scientific names, genus or family names, and the magnification power of the microscope. The established pictorial album of phytoplanktons is expected to be used by investigators for the identification of algae and the promotion of popular science education. The difficulty and time cost of algal recognition will be significantly reduced and the accuracy of identification will be greatly raised with the help of the AI-assisted algal recognition system. This innovative algal recognition technique will also become an effective tool for early-warning for accidental water quality pollution events in the future .
英文關鍵字	Phytoplankton, Biological indexes, Artificial intelligence technique

摘要

Water is the most important natural resource on the earth. The appropriate exploitation and conservation of water resources are critical to ensure social security and economic development. Therefore, stringent regulations have been promulgated through the water resource protection policies, and systematic goals have been set for management. This project including establishing a pictorial album of common algal species found in Taiwan’s water bodies and algal recognition system based on self-learning and creating AI algorithms, aims at upgrading water quality monitoring technique by using the new AI-assisted recognition technique. The first objective of this project is to establish a pictorial agal album, which includes the pictures and their characteristics of 120 phytoplankton species, and the information about their typical locations, the water environmental conditions and more. The second objective is to establish an AI-assisted recognition system for algal species. As of the mid-project stage, the system has been tested and met the preliminary training performance requirement, including the digital images of 50 algal species (genera) with confirmed scientific names, genus or family names, and the magnification power of the microscope. The established pictorial album of phytoplanktons is expected to be used by investigators for the identification of algae and the promotion of popular science education. The difficulty and time cost of algal recognition will be significantly reduced and the accuracy of identification will be greatly raised with the help of the AI-assisted algal recognition system. This innovative algal recognition technique will also become an effective tool for early-warning for accidental water quality pollution events in the future .

英文關鍵字

Phytoplankton, Biological indexes, Artificial intelligence technique