小植物，大發現 – 智慧創客顯微鏡在偏鄉小學服務
USR計畫在台灣的大學是非常重要的。這個計劃與世界各國的教育發展方向一致，也呼應聯合國SDGs的17個永續發展，而其中的如何「提昇優良教育」與「減少不平等」是我們關心的。為了達成此目標，我們開發「智慧創客顯微鏡」，帶到台灣偏鄉小學，讓孩童學習自己組出創客顯微鏡，並透過TinyML技術輔助去探索大自然。「小植物，大發現」是我們在偏鄉學校所舉辦的工作坊。在這工作坊中最大的欣慰，它帶給孩童很大的驚喜，以及學習到微觀世界的奧妙知識。
就如同許多教育資源缺乏的偏鄉小學，購買高價位顯微鏡觀是困難的。在今年的2024年4月18日，我們在台灣的一個全校只有21位學生的偏鄉沙坑(Shakeng)國小舉辦「小植物，大發現」工作坊，讓學生組裝自己的顯微鏡，並學習DIY hands-on能力；接著讓學生透過智慧創客顯微鏡探索微觀世界的奧妙知識。「智慧創客顯微鏡」結合TinyML概念與技術開發。我們使用多個tiny devices (ESP32晶片, Tiny Size Camera, 凸透鏡, 圓形OLED Display, small-size電池, micro SD卡)設計智慧創客顯微鏡的核心結構，並運用AI技術的CNN模型訓練所蒐集的資料。工作坊期間，孩子們運用智慧創客顯微鏡去發現許多新鮮事物，驚嘆聲連連地說道：「竹子的葉子長的像竹子！」，孩子們還發現沙坑校園有許多香草植物，像是：「哇！芳香萬壽菊的葉子有黃色點點，而刺蔥有白色點點。」我們的AI智慧顯微鏡系統會輔助說明，這些黃點與白點都是香草植物的腺體，香草植物透過腺體來散發香油驅趕侵犯者。
我們開發的開發智慧創客顯微鏡，目前完成 prototype設計，也完成偏鄉服務實踐。看到「小植物，大發現」工作坊帶給沙坑小學孩童驚喜與知識學習，我們感到欣慰。我們期望透過USR的實踐，提昇偏鄉小學的優良教育，並減少不平等。
我們目前開發的智慧創客顯微鏡，遭遇幾個問題：
(1)植物蒐集的Dataset目前以沙坑國小植物為主，未來需擴大。
(2)創客顯微鏡所使用的凸透鏡倍率不夠，仍須改善。
(3)植物微觀下，在顯微鏡倍率不高下看起來很接近，未來在資料訓練需做ML (Machine Learning)分群處理。
我們期望未來繼續修改我們的智慧創客顯微鏡，讓它應用至更多地方，讓「小植物，大發現」的驚喜到處發生。
The University Social Responsibility (USR) project holds significant relevance within higher education institutions in Taiwan. This initiative aligns with global educational development trends and resonates with the United Nations’ 17 Sustainable Development Goals (SDGs). Of particular concern to us are the objectives of ”Quality Education” and “Reducing Inequalities,” which are imperative to our institutional mission and global commitments.” To achieve this goal, we have developed the “Smart Maker Microscope,” which we bring to rural elementary schools in Taiwan. This initiative enables children to assemble their own maker microscopes and explore nature, assisted by TinyML technology. “Small Plants, Big Discoveries” is a workshop we conduct in rural schools. The best part of this workshop is how surprised the kids are and all the cool stuff they learn about the microscopic world.
Like many rural elementary schools suffering from a lack of educational resources, purchasing high-cost microscopes is a challenge. On April 18, 2024, we held the Small Plants, Big Discoveries workshop at Shakeng Elementary School in Taiwan, a rural school with only 21 students. The workshop allowed students to assemble their own microscopes and develop DIY hands-on skills. Following this, the students explored the wonders of the microscopic world using the Smart Maker Microscope. The Smart Maker Microscope integrates the concept and technology of TinyML. We employ multiple tiny devices, including the ESP32 chip, a tiny-sized camera, a convex lens, a round OLED display, a small-sized battery, and a micro SD card, to design the core structure of the microscope. Additionally, we utilize AI technology, specifically a CNN (Convolutional Neural Network) model, to train the data collected. During the workshop, the kids used the Smart Maker Microscope to discover all sorts of new things. They were amazed and kept saying things like, “The leaves of bamboo look just like bamboo!” They also found many aromatic plants around the Shakeng campus, such as, “Wow! The leaves of Tagetes Lemmonii (Lemmon’s marigold) have yellow spots, and those of Shi Zhu Yu (Zanthoxylum Ailanthoides) have white spots.” Our AI-powered microscope system helps explain that the yellow and white spots are the glandular tissues of the aromatic plants. These glands emit essential oils to repel invaders.
For the Smart Maker Microscope, we have completed the prototype design as well as its implementation in a rural service project. We are encouraged by the positive response and learning outcomes at the “Small Plants, Big Discoveries” workshop in Shakeng Elementary School. Through our USR project, we aim to improve education in rural schools and reduce inequality. We are currently facing several problems with the development of our Smart Maker Microscope:
(1) Currently, our plant dataset mainly includes plants from Shakeng Elementary School, and we need to add more variety in the future.
(2) The magnification of the convex lens in our maker microscope is insufficient and requires improvement.
(3) At low magnification, plants display high similarity when observed through a microscope. Consequently, ML (Machine Learning) clustering will be essential in upcoming data training efforts.
We hope to continue refining our Smart Maker Microscope in the future, expanding its application to more areas, and allowing the “Small Plants, Big Discoveries” surprises to occur everywhere.
Keywords: University Social Responnsibility, Microscope, TinyML, Rural Elementary Schools, Aromatic Plants, Machine Learning.
