Review Article 評論文章
Applications of artificial intelligence in sericulture
人工智慧在養蠶業中的應用
Abstract 抽象
The Sericulture industry is one of the major cottage industries producing higher income with lower input. The industry requires critical inputs like quality seed, quality feed, skilled labour with optimum environmental conditions for smooth running and higher production. Most of these input processes involve only manual assesment of phenotypic traits and are human-centric. In the current environment, producing high-quality silk is crucial to reaching sustainability by 2030 . The identification of the barriers preventing the increase of silkworm production is limited by the traditional technique. With the development of artificial intelligence, it is providing many benefits to sectors like sericulture where expert systems are being used to solve many problems like disease and pest, gender classification, changing environmental conditions in both host plant as well as silkworm. For the sericulture industry to thrive amidst a changing world, it must keep pace with evolving challenges. This necessitates emphasizing the integration of intelligent tools. There are various advanced tools like Artificial Intelligence, advanced mechanisations etc. and their use is limited to some extent. As the use of Al is gaining momentum, the sericulture industry is also bound to use them to some extent even though they are not very much popular. This article underscores the importance of leveraging technology for a prosperous future and economy.
養蠶業是主要的家庭手工業之一，以較低的投入產生更高的收入。該行業需要關鍵的投入，如優質種子、優質飼料、熟練的工作力以及最佳的環境條件，以實現平穩運行和更高的產量。這些輸入過程中的大多數僅涉及表型性狀的手動評估，並且以人類為中心。在當前環境下，生產高品質的絲綢對於到 2030 年實現可持續發展至關重要。傳統技術對阻礙養蠶產量增加的障礙的識別受到限制。隨著人工智慧的發展，它為養蠶業等行業提供了許多好處，其中專家系統被用於解決許多問題，例如病蟲害、性別分類、寄主植物和蠶不斷變化的環境條件。養蠶業要在不斷變化的世界中蓬勃發展，必須跟上不斷變化的挑戰。這需要強調智慧工具的集成。有各種高級工具，如人工智慧、高級機械化等，它們的使用在一定程度上受到限制。隨著Al的使用勢頭越來越大，養蠶業也必然會在一定程度上使用它們，即使它們不是很受歡迎。本文強調了利用技術促進繁榮未來和經濟的重要性。
Key words: Artificial Intelligence, voice recognition, image processing, Sericulture, pest and disease, gender classification, host plant
關鍵詞 ： 人工智慧， 語音辨識， 圖像處理， 養蠶， 病蟲害， 性別分類， 寄主植物
1. INTRODUCTION 1. 引言
The English word “Sericulture” came from the Greek word “Sericos”, meaning silk and English word “culture” i.e. rearing, which entails the complete process of cultivating silkworms, growing their food plants, and producing silk (Choudhury et al., 2020). This agro-industry is economically beneficial and integrates various activities that have a profound impact on the economic prosperity of rural areas. Sericulture merges agricultural activities like silkworm rearing and food plant cultivation with industrial methods for silk production, which includes breeding silkworms to obtain eggs and cocoons. In India, sericulture i.e. the rearing of silkworms and subsequent production of silk fiber, has become a promising rural industry. This is attributed to its short gestation period, low initial
英文單詞“Sericulture”來自希臘語“Sericos”，意思是絲綢和英語單詞“culture”，即飼養，它涉及養蠶、種植食用植物和生產絲綢的完整過程（Choudhury 等人，2020 年）。這種農業工業在經濟上是有益的，並整合了對農村地區經濟繁榮產生深遠影響的各種活動。養蠶業將養蠶和食用植物種植等農業活動與工業製絲方法相結合，其中包括飼養蠶以獲得卵和繭。在印度，養蠶業，即養蠶和隨後的絲纖維生產，已成為一個有前途的農村產業。這歸因於其妊娠期短、初始時間短
costs, significant employment opportunities, and high potential for returns on investment. Over the course of time, this industry is evolving to meet the demands of the market, new advances in technology, and concerns for the environment. The raw silk production statistics for the year of 2022-2023 was 36,582 MT while there was an increase in the production statistics for the year of 2023-2024 i.e. 38,913 MT (Source: CSB, Bengaluru). However, currently sericulture has become less about making silk traditionally and more about adopting innovative technologies, ecological methods, and responding to the needs of a worldwide market. Sustainable sericulture is more important than ever as environmental awareness expands.
成本、大量就業機會和高投資回報潛力。隨著時間的推移，該行業不斷發展以滿足市場需求、技術的新進步和對環境的關注。2022-2023 年的生絲產量統計數據為36,582噸，而2023-2024年的產量統計數據有所增加，即38,913噸（來源：CSB，班加羅爾）。然而，目前養蠶業已經不再是傳統上製作絲綢，而是更多地採用創新技術、生態方法和回應全球市場的需求。隨著環保意識的擴大，可持續養蠶比以往任何時候都更加重要。
The industrial process of producing silk is changing these days due to the integration of cutting-edge technologies including cloud computing, big data, artificial intelligence (AI), and the Internet of Things (IoT). In sericulture, production capacity undoubtedly depends on healthy and nutritious food plants which is the primary requirement. Food plant improvement includes proper nutrient management, protection from diseases and pests etc. Here, AI provides one of the most appealing technologies i.e. disease detection using Machine learning algorithms such as Convolutional neural networks and support vector machines which demonstrate exceptional proficiency in categorizing photos of plants and their leaves to identify visual indications of disease. The manual inspection is time consuming, while human error is one of the limiting factors. Hence, use of Al significantly improves accuracy while reducing time consumption (Vijayreddy, 2023). Use of Deep Learning methods such as CNN in disease and pest detection is also being widely accepted in the field of agriculture (Liu and Wang, 2021). Side by side, availability of healthy seed production is also of top priority. Counting small objects like silkworm eggs manually becomes challenging when they overlap or appear in large numbers, leading to errors and consuming a lot of time. Computer vision is necessary to replace manual methods due to their inaccuracies and inefficiencies (Pathan and Harale, 2016).
由於雲計算、大數據、人工智慧 （AI） 和物聯網 （IoT） 等尖端技術的整合，如今生產絲綢的工業流程正在發生變化。在養蠶業中，生產能力無疑取決於健康和營養的食用植物，這是首要要求。食品植物的改進包括適當的營養管理、預防病蟲害等。在這裡，人工智慧提供了最吸引人的技術之一，即使用機器學習演算法（如卷積神經網路和支援向量機）進行疾病檢測，這些演算法在對植物及其葉子的照片進行分類以識別疾病的視覺跡象方面表現出非凡的能力。人工檢查很耗時，而人為錯誤是限制因素之一。因此，使用 Al 可以顯著提高準確性，同時減少時間消耗（Vijayreddy，2023 年）。在疾病和害蟲檢測中使用CNN等深度學習方法在農業領域也被廣泛接受（Liu和 Wang，2021年）。與此同時，健康種子生產的可用性也是重中之重。當蠶卵等小物體重疊或大量出現時，手動計數變得具有挑戰性，從而導致錯誤並消耗大量時間。由於手動方法的不準確和效率低下，計算機視覺是取代手動方法所必需的（Pathan 和 Harale，2016 年）。
2. ARTIFICIAL INTELLIGENCE
2. 人工智慧
Artificial Intelligence defined by Grewal (2014) as “The mechanical simulation system of collecting knowledge and information and processing intelligence of the universe: (collecting and interpreting) and disseminating it to the eligible in the form of actionable intelligence”. Simply put, artificial intelligence refers to machines’ capacity to emulate human behavior. These machines can think, learn, and act like humans, applying their knowledge in real-time to solve problems. The well-known examples of artificial intelligence are iPhone Siri, amazon Alexa, IBMs Deep Blue, Sophia the humanoid robot etc.
人工智慧被 Grewal （2014） 定義為「收集知識和資訊並處理宇宙智慧的機械模擬系統：（收集和解釋）並以可作情報的形式將其傳播給符合條件的人」。。簡而言之，人工智慧是指機器類比人類行為的能力。這些機器可以像人類一樣思考、學習和行動，即時應用他們的知識來解決問題。人工智慧的著名例子是iPhone Siri、亞馬遜Alexa、IBM Deep Blue、人形機器人 Sophia 等。
MACHINE LEARNING 機器學習
Machine learning fundamentally revolves around the idea of learning from experience and examples, forming a subset of artificial intelligence. It employs algorithms and statistical techniques that enable machines to learn from data, generate results based on their training and testing, and progressively improve their performance as they encounter more data and experience. In the same way that a child is taught to walk, talk,
機器學習從根本上圍繞著從經驗和示例中學習的理念展開，形成了人工智慧的一個子集。它採用演算法和統計技術，使機器能夠從數據中學習，根據訓練和測試生成結果，並在遇到更多數據和經驗時逐步提高其性能。就像教孩子走路、說話、
run, read, and write, a computer is taught everything to become artificially intelligent through machine learning, language, and commands. Machine learning makes use of several technologies such as computer vision, robotics, image processing, and symbolic learning. The computer automatically evaluates tens of thousands of samples, generates an algorithm, and then enhances itself through machine learning once it reaches the intended result. Human inputs are required for the machine to learn. The machine is taught by the human that for this specific activity, you should do this and that its output should be what we desire, and then the human leaves the machine to learn automatically from its prior experiences. Therefore, the fundamental data bases and the cycles or processes used to fill these databases into the machines should be reliable, secure, and up to date.The qualities of artificial or machine intelligence include mobility, understanding, forecasting modification, spontaneous decision-making, and constant learning (Mohammed, 2019).
運行、讀取和寫入，計算機通過機器學習、語言和命令教會一切變得人工智慧。機器學習利用了多種技術，例如計算機視覺、機器人技術、圖像處理和符號學習。計算機會自動評估數以萬計的樣本，生成演算法，然後在達到預期結果後通過機器學習進行自我增強。機器需要人工輸入才能學習。機器被人類教導，對於這個特定的活動，你應該做這個，它的輸出應該是我們想要的，然後人類離開機器，自動從它以前的經驗中學習。因此，用於將這些資料庫填充到機器中的基本資料庫和週期或過程應該是可靠、安全和最新的。人工智慧或機器智慧的品質包括移動性、理解力、預測修改、自發決策和持續學習（Mohammed，2019 年）。
Al technologies are being used in various fields of agriculture. Machine Learning and AI in agriculture aim to achieve precision farming, minimizing natural resource depletion and waste. They predict weather for optimized fertilizer use and model crops to forecast yield, pest and disease risks, and soil conditions, offering strategic recommendations (Pal et al., 2023). Use of drones incorporated with artificial intelligence technologies like thermal, multispectral and hyperspectral sensors is gaining importance as it can acquire vast areas of agricultural land for monitoring various biotic and abiotic factors that will ultimately result in increased productivity while consuming less time and labour (Slimani et al., 2023). Remote sensing has also become an esteemed tool for farmers as well as agriculturists for monitoring crop health, soil nutrient management, estimate growth & yield index, overall crop management strategies etc. (Reddy Vijayreddy, 2023).
Al 技術正在農業的各個領域得到應用。農業中的機器學習和 AI 旨在實現精準農業，最大限度地減少自然資源的消耗和浪費。他們預測天氣以優化肥料使用，並對作物進行建模以預測產量、病蟲害風險和土壤條件，從而提供戰略建議（Pal et al.， 2023）。使用與熱、多光譜和高光譜感測器等人工智慧技術相結合的無人機變得越來越重要，因為它可以獲得大面積的農業用地來監測各種生物和非生物因素，最終提高生產力，同時消耗更少的時間和工作力（Slimani 等人，2023 年）。遙感也已成為農民和農業學家監測作物健康、土壤養分管理、估計生長和產量指數、整體作物管理策略等的受人尊敬的工具（Reddy Vijayreddy， 2023）。
Issues facing the sericulture industry include preserving humidity and temperature, reducing the number of workers needed to count eggs, and separating sex, among others. Constant counting can cause vertigo, migraines, and eye strain in addition to reducing the precision of the results. In order to overcome these challenges in the field of sericulture today, computer-based approaches such as Artificial Neural Networks (ANN), Single Shot Multibox Detector (SSD), Convolutional Neural Network (CNN), the Internet of Things (IoT), Artificial Intelligence (Al) methodologies, and image processing algorithms must be employed. In the modern sericulture industry, computer-based technologies are significantly more important to reaching tangible goals. It’s also used in the sericulture industry for counting, identifying, and classifying fruits, flowers, leaves, sicknesses, and silkworm eggs, among other
things.In the past, researchers have used deep learning (VGG16, ResNet50, and Inception V3), machine learning (ANN, CNN, Fast CNN, Faster CNN, SSD, Yolo V3, KNN, SVM), and image processing (Violo) techniques to efficiently count, detect, and classify silkworm eggs at low cost. This is because early detection of unhealthy and healthy silkworm eggs leads to an estimated
profit.An Agritech start-up in Bengaluru is connecting the entire silk industry supply chain with digital technologies, enabling sericulture farmers to get better prices for their produce and guaranteeing the quality of cocoons and yarn for reelers, weavers, and retailers.A range of techniques, such as human approaches, artificial neural networks (ANN), the internet of things (IOT),
養蠶業面臨的問題包括保持濕度和溫度、減少計數雞蛋所需的工人數量以及區分性別等。不斷計數除了會降低結果的精度外，還會導致眩暈、偏頭痛和眼睛疲勞。為了克服當今養蠶領域的這些挑戰，必須採用基於計算機的方法，例如人工神經網路 （ANN）、單發多箱檢測器 （SSD）、卷積神經網路 （CNN）、物聯網 （IoT）、人工智慧 （Al） 方法和圖像處理演算法。在現代養蠶業中，基於計算機的技術對於實現有形目標的重要性要大得多。它還用於養蠶業，用於對水果、花、葉、病和蠶卵等進行計數、識別和分類。過去，研究人員使用深度學習（VGG16、ResNet50 和 Inception V3）、機器學習（ANN、CNN、Fast CNN、Faster CNN、SSD、Yolo V3、KNN、SVM）和圖像處理 （Violo） 技術來高效計數、檢測和分類蠶卵，成本低。這是因為早期發現不健康和健康的蠶卵會帶來估計的利潤。班加羅爾的一家農業科技初創公司正在利用數位技術連接整個絲綢行業供應鏈，使養蠶農民能夠獲得更好的產品價格，並保證卷筒機、織工和零售商的蠶繭和紗線的品質。一系列技術，例如人類方法、人工神經網路 （ANN）、物聯網 （IOT）、