頁籤選單縮合
| 題 名 | 具備深度學習之愛文芒果等級區分影像辨識無人機=Aiwen Mango Grade Classification by UAV with Deep Learning Image Recognition |
|---|---|
| 作 者 | 黃佳信; 許弘維; 吳亦超; | 書刊名 | 臺東大學綠色科學學刊 |
| 卷 期 | 14:2 2024.11[民113.11] |
| 頁 次 | 頁36-52 |
| 分類號 | 312.831 |
| 關鍵詞 | 愛文芒果; 芒果等級辨識; 深度學習; 影像辨識; 無人機; Aiwen mango; Grade classification; Image recognition; UAV; YOLO; Deep learning; |
| 語 文 | 中文(Chinese) |
| 中文摘要 | 水果一直是臺灣農產品外銷的重大主力,其中愛文芒果更是外銷水果中之一大主 力。隨著臺灣人口高齡化,加上投入農業產業活動的人口較為稀少,導致需要大量人 力支援的生產、包裝與運送成本都大幅提高。在包裝的過程中,除了需要大量人力外, 當中的芒果等級辨識更需要繁瑣細心的工作流程。因此芒果等級辨識已成為農民們除 了品種改良及行銷外,最為煩惱的問題。為此,許多研究提出結合 AI 影像辨識方式 來解決此問題。然而目前的 AI 芒果影像等級辨識仍多以手持攝像裝置來完成芒果等 級辨識,並無法適用於果園內之芒果等級辨識。有鑑於此,本論文結合 UAV 與 AI 影 像辨識,提出「具備深度學習之愛文芒果等級區分影像辨識無人機」,透過 UAV 在不 用跑遍果園的情況下,便能確實辨識樹上芒果的等級,並透過真實 UAV 攝像方式, 來驗證本論文所提出方法的有效性。透過實驗結果,本論文在愛文芒果等級辨識可達 到 70%,證明本論文提出之「具備深度學習之愛文芒果等級區分影像辨識無人機」可 適用於愛文芒果等級區分,並可降低大量人力成本。 |
| 英文摘要 | In Taiwan, fruits are the major contribution of agricultural exports. Aiwen mango is one of the main export fruits of agricultural exports. Since the population involved in agricultural industry activities is relatively small due to aging population in Taiwan, the cost of production, packaging and shipping with a lot of manpower support is increased significantly. As of 2023, Taiwan’s mango plantation area is about 16,000 hectares with an annual output of about 167,000 metric tons. However, the population engaged in agricultural industrial activities is relatively sparse with a high age group by mostly manual harvesting and quality screening to lead to a lack of more convenient and standardized processes. Therefore, an Aiwen mango grade classification by image recognition UAV with YOLO deep learning, AMGCIR, was proposed to address the above issues. In AMGCIR, the UAV could be operated with a laptop to detect the mango's grade without running through the orchard and the manual quality screening in orchard before mango harvest. The experimental results showed that the accurate rate of grade classification could be over 70%. It proved that AMGCIR could be used for Aiwen mango grade classification to reduce the manpower and avoid the manual quality screening in orchard before mango harvest. The export and transportation of high-grade fruits thus could be arranged in advance. |
本系統中英文摘要資訊取自各篇刊載內容。