| dc.contributor.advisor | 蔡子傑 | zh_TW |
| dc.contributor.advisor | Tsai, Tzu-Chieh | en_US |
| dc.contributor.author (Authors) | 劉敏傑 | zh_TW |
| dc.contributor.author (Authors) | Liu, Min-Chieh | en_US |
| dc.creator (作者) | 劉敏傑 | zh_TW |
| dc.creator (作者) | Liu, Min-Chieh | en_US |
| dc.date (日期) | 2021 | en_US |
| dc.date.accessioned | 1-Nov-2021 12:19:12 (UTC+8) | - |
| dc.date.available | 1-Nov-2021 12:19:12 (UTC+8) | - |
| dc.date.issued (上傳時間) | 1-Nov-2021 12:19:12 (UTC+8) | - |
| dc.identifier (Other Identifiers) | G0102971009 | en_US |
| dc.identifier.uri (URI) | http://nccur.lib.nccu.edu.tw/handle/140.119/137730 | - |
| dc.description (描述) | 碩士 | zh_TW |
| dc.description (描述) | 國立政治大學 | zh_TW |
| dc.description (描述) | 資訊科學系碩士在職專班 | zh_TW |
| dc.description (描述) | 102971009 | zh_TW |
| dc.description.abstract (摘要) | 近年來，許多ICT業者開始積極於物聯網產業尋找新藍海，其中智慧穿戴式裝置因具有解放雙手及隨時量測之優勢，並隨著Apple公司與各大廠爭相投入，整體產業邁入成長期。但他們大部將此應用於生理健康方面，較少針對特定的運動項目開發各式的專業應用裝置。為此，本論文希望設計輕便型穿戴裝置，針對羽毛球運動訓練過程，可即時傳回感測器的資料，搭配AI深度學習的技術，開發一套輔助訓練系統。本論文所建構的軟硬體系統均採用開放式架構來建置，增加了整個系統的開發彈性、相容性、以及可擴充性。使用輕便型手腕感測套，而不用嵌入在球拍內，增加方便性。所收集的感測器資料取自於校隊球員揮拍訓練，並針對揮拍擊球過程進行分析。首先，我們建構斷拍演算法，以使能有效擷取球員每一次揮拍過程的數據；接著進行AI深度學習，以CNN(Convolutional Neural Networks)與LSTM(Long short-term memory)演算法，判別拍種預測，分別可得到高達96.74%以及97.83%的準確率。我們更進一步建構球員的等級預測模型，分別得到70.27%以及80.63%的結果。另外我們也初步建構單一揮拍評分模型，以供球員及教練評估該次揮拍的狀況。期許本研究提出之系統架構與方法只是一個開始，未來可讓更多穿戴裝置應用於其他專業運動訓練領域。 | zh_TW |
| dc.description.abstract (摘要) | Recently, many ICT companies have begun to actively seek a new blue ocean in the Internet of Things industry. With the advantages of freeing hands and measuring at any time, many major manufacturers such as Apple are developing the smart wearable devices. The entire industry in this field has entered a growth stage. However, most of them are applied to biosensors for health, and just a few are for professional specific sports. In this thesis, we design a lightweight wearable wristlet with sensors that can transmit the sensed data in real time during the badminton training course. We then apply to the AI deep learning technologies, and integrate this into an auxiliary training system for badminton players.The software and hardware systems constructed in this thesis are all built with an open architecture, and thus increase the development flexibility, compatibility, and expandability. The lightweight wristlet design, instead of embedded sensors into racket grips, increases the convenience and comfortability. The data are collected from the school badminton team, and the swing trajectory is analyzed. First, we construct an algorithm for segmenting each swing in the entire training course. Then, we apply the AI deep learning algorithms, CNN (Convolutional Neural Networks) and LSTM (Long short-term memory), to these swing data for constructing the swing type models. The test results can get up to 96.74% and 97.83% accuracy, respectively. We further label the technical levels for each player by the coach. We construct the scoring models, and the test results can get 70.27% and 80.63% accuracy, respectively. In addition, it is possible for any swing to be evaluated for a given swing type for more detailed assessments. This study is just a beginning, and we wish it can be extended to more other professional sports training courses in the future. | en_US |
| dc.description.tableofcontents | 第一章 簡介 11.1 背景 11.2 動機 21.3 研究目的 2第二章 相關研究 42.1 硬體架構 42.1.1 Arduino 42.1.2 Raspberry Pi樹莓派開發板 62.1.3 低功耗藍牙BLE傳輸器 72.1.4 姿態感測器 82.2 深度學習架構 8第三章 硬體擴充 93.1 Arduino電源的選用 93.1.1 5V消費性行動電源 103.1.2 3.7V Li-Po(Lithium Polymer)鋰電池 103.1.3 3.7V 18650鋰電池 123.1.4水銀電池 123.2評估數據FPS與準確性 153.2.1 IMU資料接收 163.2.2 藍芽資料接收 17第四章 系統架構 204.1 資料取得 214.1.1 資料來源 214.1.2 讀取資料 264.2 資料前處理 274.2.1 視覺化展現 274.2.2 加入變化量數值 284.2.3 擷取單一揮拍資料 304.3 演算法選用 324.3.1 CNN(Convolutional Neural Networks)演算法 324.3.2 LSTM(Long short-term memory)演算法 344.3.3 KNN(k-nearest neighbors)演算法 364.4 資料格式化 384.4.1 單一揮拍數據資料集 384.4.2 擊球點中心數據資料集 39第五章 實驗結果 415.1 資料匯入 415.2 CNN演算法 425.3 LSTM演算法 465.4 LSTM評分法 495.5 KNN演算法 51第六章 結論與未來展望 536.1 結論 536.2 未來展望 53參考資料 55 | zh_TW |
| dc.format.extent | 3134322 bytes | - |
| dc.format.mimetype | application/pdf | - |
| dc.source.uri (資料來源) | http://thesis.lib.nccu.edu.tw/record/#G0102971009 | en_US |
| dc.subject (關鍵詞) | 智慧穿戴裝置 | zh_TW |
| dc.subject (關鍵詞) | 輕便型手腕感測套 | zh_TW |
| dc.subject (關鍵詞) | 深度學習 | zh_TW |
| dc.subject (關鍵詞) | CNN | zh_TW |
| dc.subject (關鍵詞) | LSTM | zh_TW |
| dc.subject (關鍵詞) | 羽毛球輔助訓練系統 | zh_TW |
| dc.subject (關鍵詞) | Smart wearable devices | en_US |
| dc.subject (關鍵詞) | Light weight wristlet | en_US |
| dc.subject (關鍵詞) | Deep learning | en_US |
| dc.subject (關鍵詞) | CNN | en_US |
| dc.subject (關鍵詞) | LSTM | en_US |
| dc.subject (關鍵詞) | Badminton auxiliary training system | en_US |
| dc.title (題名) | 以輕便型手腕感測套之智慧型羽球揮拍動作辨識與評分系統 | zh_TW |
| dc.title (題名) | A Smart Badminton Swing Recognition & Scoring System based on Lightweight Wrist Sensors | en_US |
| dc.type (資料類型) | thesis | en_US |
| dc.relation.reference (參考文獻) | [1] 黃聖智, 基於加速度特徵值之模糊手勢識別系統. International Journal of Advanced Information Technologies (IJAIT), Vol. 7, No.2.[2] "羽球-維基百科". https://zh.wikipedia.org/wiki/%E7%BE%BD%E6%AF%9B%E7%90%83[3] Indrajeet Ghosh; Sreenivasan Ramasamy Ramamurthy; Nirmalya Roy: “StanceScorer: A Data Driven Approach to Score Badminton Player”: IEEE International Conference on Pervasive Computing and Communications Workshops, 2020.[4] 蘇冠榮, 穿戴式互動展演創新應用與姿態感測技術研究. Jan 2016.[5] Wenjing Su; Jiang Zhu; Huan Liao; Manos M. Tentzeris: “Wearable Antennas for Cross-Body Communication and Human Activity Recognition”: IEEE Access (Volume: 8), 2020.[6] Arduino, https://www.arduino.cc/[7] Arduino Nano 33 BLE Sense, https://store-usa.arduino.cc/products/arduino-nano-33-ble-sense[8] Raspberry Pi, https://www.raspberrypi.org/[9] Specification of the Bluetooth System, Covered Core Package, Version: 5.0; The Bluetooth Special Interest Group: Kirkland, WA, USA, 2016.[10] Pierluigi Casale; Oriol Pujol; Petia Radeva: ” Human Activity Recognition from Accelerometer Data Using a Wearable Device”: Lecture Notes in Computer Science book series(LNCS, volume 6669), 2011.[11] Guto Leoni Santos; Patricia Takako Endo; Kayo Henrique de Carvalho Monteiro: “Accelerometer-Based Human Fall Detection Using Convolutional Neural Networks”: Sensors (Basel), 2019.[12] WISDM: Wireless Sensor Data Mining, https://www.cis.fordham.edu/wisdm/dataset.php[13] Smashing a world record, again - NANORAY Z-SPEED, https://www.youtube.com/watch?v=uhWkVYqo_bo[14] Femme Verbeek, Version 2.0 of the LSM9DS1 library, https://github.com/FemmeVerbeek/Arduino_LSM9DS1[15] Yann LeCun; Leon Bottou; Yoshua Bengio; Patrick Haffner, “Gradient-Based Learning Applied to Document Recognition”, PROC OF THE IEEE, Nov 1998. | zh_TW |
| dc.identifier.doi (DOI) | 10.6814/NCCU202101687 | en_US |
