瓢蟲的自動化雌雄分離系統應用於六條瓢蟲

Abstract

隨著食品安全和環保意識的提升，農藥減量政策成為全世界農業發展的重要方向，提升了生物防治技術的關注度。六條瓢蟲為鞘翅目瓢蟲科，以蚜蟲為主食，為農業害蟲防治的重要天敵昆蟲，在生物防治中有卓越表現。本研究針對六條瓢蟲（Cheilomenes sexmaculata），開發了一套自動化成蟲雌雄分離系統。然而，現有的大量瓢蟲養殖方式，無法做到雌雄分離。雌雄分離在瓢蟲的生產中具有重要意義。雌蟲的主要任務是產卵，繁衍下一代。雄蟲主要負責交配，而在高密度環境中雄蟲之間會產生競爭關係，影響繁殖效率。而通過雌雄分離，可以調整雌雄比，確保繁殖成功率也預防了競爭問題。除此之外也能防止雄蟲食卵的問題，從而提升整體生產效率。所以在無分辨雌雄的情況下難以達到量產，且分辨瓢蟲雌雄需依賴大量人力，難以有效提升生產效率。 為了解決此問題，本研究開發自動化瓢蟲成蟲雌雄分離系統。此系統的目標是解決在大規模瓢蟲飼養過程中，雌雄分離所涉及的人力和耗時問題。本研究欲透過機器視覺技術，精準的定位瓢蟲的位置並辨識其性別。此系統結合六軸移動平台與末端吸取裝置，在吸取過程中實現分類，確保瓢蟲被正確歸類。本研究建立之自動化系統可達成以下三個目標：1. 建立具備雌雄分離功能之六軸移動平台；2. 建立YOLOv8深度學習模型精準定位瓢蟲中心點位置；3. 以YOLOv8建立雌雄辨識模型。結果顯示，在精準定位方面，精準定位後的吸取成功率可達94.6%，系統所建立的雌雄辨識模型在測試集中準確度達到91.8%；Precision在雌性瓢蟲高達97.8%，雄性瓢蟲達到87.8%；Recall在雌性瓢蟲上達88.7 %，雄性瓢蟲高達98.8%；F1-score在雌性瓢蟲達93.1 %，雄性瓢蟲達92.9 %。而在實際辨識雌雄瓢蟲時可達92%的準確率，而辨識後之放置成功率高達97.8%，表示系統能夠精確且穩定的完成瓢蟲的雌雄分離工作。該自動化系統不僅減少人工操作需求，降低人力成本，還能提高生產效率，為瓢蟲大規模飼養提供了一個可行的解決方法。此系統的成功開發將促進生物防治技術的發展，為瓢蟲養殖業提供實際應用的技術支持。

With the increasing awareness of food safety and environmental protection, the policy of reducing pesticide usage has become a significant direction for global agricultural development, enhancing the focus on biological control techniques. The Cheilomenes sexmaculata is the Coccinellidae family in the order Coleoptera, primarily feeds on aphids and is an important natural enemy in agricultural pest control, demonstrating outstanding performance in biological control. This study aims to develop an automated gender separate system for ladybug. However, the current large-scale rearing methods for ladybugs present challenges for effective gender separation. Gender separation is crucial in ladybug production, as female ladybugs are primarily responsible for laying eggs and propagating the next generation, while males are mainly involved in mating. In high-density environments, competition among males can affect reproductive efficiency. By separating males and females, the gender ratio can be adjusted to ensure reproductive success and prevent competition. Additionally, it can prevent males from eating eggs, thereby improving overall production efficiency. Without gender separation, it is challenging to achieve mass production, and gender identification of ladybugs requires substantial manual labor, making it difficult to enhance production efficiency effectively. To address this issue, this study developed an automated system for gender separation of adult ladybugs. The goal of this system is to address the labor and time-consuming challenges associated with gender separation in large-scale ladybug rearing. The system aims to accurately locate ladybugs and identify their gender through machine vision technology. This system combines a six-axis moving platform with an end effector to achieve classification during the suction process, ensuring that ladybugs are correctly categorized. The automated system developed in this study can achieve the following three objectives: 1. Establish a six-axis moving platform with gender separation capability; 2. Develop a YOLOv8 deep learning model to accurately locate the central point of ladybugs; 3. Develop a YOLOv8 model for gender identification. The results show that the success rate of suction after accurate positioning can reach 94.6%. The gender identification model developed by the system achieves an accuracy of 91.8% in the test set, with a precision of 97.8% for females and 87.8% for males, a recall of 88.7% for females and 98.8% for males, and an F1-score of 93.1% for females and 92.9% for males. In practical applications, the system can achieve an accuracy rate of 92% for gender identification, and a placement success rate of up to 97.8%, indicating that the system can accurately and reliably complete the gender separation of ladybugs. This automated system not only reduces the need for manual operation and lowers labor costs but also improves production efficiency, providing a feasible solution for large-scale ladybug rearing. The successful development of this system will promote the advancement of biological control technology and provide practical technical support for the ladybug rearing industry.