Zabnbzf2ltrt9qvrrkph

杜威慶

我是杜威慶,畢業於國立臺南大學資訊工程研究所,擔任軟體工程師。 負責在微軟平台上的Corel公司的大型修圖軟體,研發與維護的工作,雖然自己專長是IoT方面,但在工作上不會因此受限,反而會與同事或主管一起研究找出適合的圖像處理演算法再透由OpenCV去實現。對於學習新的知識有著強烈的企圖心與積極的心。

工作中大量使用Python、C/C++參與多項Paint shop新功能開發,也能快速配合客戶的需求,不侷限於RD的職位,會與QA合作與溝通,盡可能減少side effect。我也會利用空閒時間進修新的知識理論,讓自己更快適應日新月異的科技環境,並樂於分享使自己和團隊更進步。
My name is WEI-QING, DU. I graduated from the Institute of Computer Science, National Tainan University. I am currently working at Ulead Digital Technology as a software engineer. 

I responsible for the research, development and maintenance of Corel’s  application software on the Microsoft platform. Although my expertise is in Internet of Things(IoT), I don't put limits on myself. 

The image processing algorithm is then implemented by OpenCV. Have a strong ambition and positive heart for learning new knowledge. I use Python and C/C++ extensively in my work to participate to develop features in the PaintShop Pro(PSP) . Find suitable algorithms to satisfy customer's request. I also cooperating and communicating with QA to minimize side effects. 

軟體工程師 Software Engineer
TNN,TW
工作相關邀約請聯繫:[email protected]
固定參與社群:
Taiwan 程式語言讀書會

KSDG - 高雄軟體開發者社群



學歷/ Education

國立臺南大學(National University of Tainan), Sep 2015 - Jul 2017

科系:資訊工程研究所 Institute of Computer Science。 

學歷:碩士 Master's degree

碩士論文: Energy- Efficient Data Collection Using Mobile Robot in Wireless Sensor Networks (WSN)。

Experiences 00 00@2x 504900dc09d82f711fdb54cf5763251cab5193a844856da978b2b8f3907ecd5a

崑山科技大學(Kun Shan University) ,Sep 2011 - Jun 2015

科系:電腦與通訊系 Department of Computer and Communication
學歷:學士 Bachelor
專題製作: 心電圖量測系統。


Experiences 00 01@2x 6daa47d1df70d315f129dcecdd562003831581c4cdd8f9c819aa22d4a24ae150

經歷 / Experience

友立數位科技有限股份公司 Ulictek, Oct 2018 - now

前身為製作photo impact軟體的友立資訊,目前提供專業的圖像處理的服務。
Ulictek used to be a company that produced photo impact software, and currently provides professional image processing services.
  • 大量使用OpenCV,節省60%的開發時間。
  • Use OpenCV extensively, saving 60% of development time.
  • 我目前負責開發與維護Corel公司的大型修圖軟體 Paint shop (PSP)2020。 
  • I am currently responsible for the development and maintenance of Corel's  software Paint shop (PSP) 2020.
  • 在半年時間合作開發新功能Refine Brush,並且建立新的UI。 
  • In half a year, we cooperated to develop the new feature of Refine Brush and build a new UI.
  • Refine Brush方便使用者,在去除背景時,同時連細微髮絲,都能夠選取的到。
  • Refine Brush is convenient for users when they removing the background even the fine hair can be selected.
  •  Refine Brush介紹影片。
  • 開發AI的方式處理有雜訊的圖片。(Super Resolution Deep Learning)
  • Developed AI to process noisy pictures.


科技部計畫 (Ministry of Science and Technology): 無線與可充電感測網路中設計與實作能源效率移動機器人資料收集與無線充電策略,Aug 2017

  • 在研討會獲得報告的機會後,再將無線充電議題融入到自己研究中,並撰寫科技部計畫書。最後獲得科技部計畫經費補助,也讓自己的研究成果具有發展潛力。 科技部計畫編號: MOST-106-2221-E-024 -002 -MY2。

獲第十三屆無線、隨意及感測網路研討會議接受為口頭報告發表,Aug 2017

  •   專注於Robot路徑研究,如何讓路竟能更短,將自己幾番改進、研讀、實作制做成論文,並投稿於研討會。榮幸能獲得主辦單位認同,有機會在會議中報告給師長與先進前輩聽,從中獲得不錯的建議。

書卷獎,Sep 2011 - Jun 2015

  • 因為大學期間,懂得分配讀書時間,使得自己成績優異,每學期都獲得書卷獎的殊榮,自己的努力受到認同與獎勵。

大學專題-心電圖量測系統,Mar 2014 – Jun 2015

  • 由於現今人們生活忙碌,各種壓力與疲勞造成身體上疾病的發生。  然而以心臟方面疾病更是長年高居國人十大死因前幾名,經此專題製作開發居家便利使用的簡易心臟之心電圖量測裝置。  本專題開發一套簡易型、低價位生理訊號量測模組,採用類比訊號測量電路、MSP430 微處理器、藍牙無線傳輸模組所組成,可進行簡易之心電圖量測系統。

全國LED創意設計競賽-決選入選,Feb 2013

  • 參賽組別:應用產設計組。
  • 作品名稱:彩光。
  • 以向日葵為外型,一開始製作LED的電路板,配合Microchip Technology的開發板連結紅外線感測器,使得LED有不同的變化。

技能 Skills


C/C++

開發TI MSP430、Microchip

Developed TI MSP430、Microchip.

自走車的API

API for self-propelled vehicles.

Windows大型軟體開發。

Windows large-scale software development.

微軟基礎類別庫,Microsoft Foundation Classes (MFC)。


Front-end

-Corel Paintshop 2021 UI建置

-HTML

-Angular

-Javascript




C#

-視窗設計

-Window design
-UI介面設計

-UI interface design


開發工具 Development tools

-JetBrains 

-Eclipse 

-VS (Visual Studio)

-VScode


Python

-Web Crawler.

-Test the algorithm.

-Evaluate the performance.


Tools

-Git
-Sourcetree
-Fork

Domain Driven Design(Studying)

-程式碼即是文件。

The code is the document.

-加強與非技術人員的溝通。

Strengthen communication with non-technical personnel.

-加速團隊的開發。

Accelerate team development.




專案經歷 Project experience


Electrocardiography(ECG)

-使用開發板TI MSP430

Use the development board TI MSP430.

-撰寫C/C++

Compose C/C++.

-擷取人體的心電訊號

Capture the body's ECG signal.


手機繪製心電圖

-透過藍芽模組,將心電訊號傳至手機中

Transmit the ECG signal to the mobile phone through the Bluetooth module.
-撰寫Android將訊號繪製出來

Compose Android to draw the signal.


Arduino IoT BOE-BOT (ABB Car)

-使用Arduino IoT BOE-BOT (ABB Car)

Use Arduino IoT BOE-BOT (ABB Car)

-撰寫C/C++

Compose C/C++.

-撰寫車子向前、左、右、後的API

Compose the API of the car forward, left, right and back.


Looking for the scooter with smartphone based on BLE (  Bluetooth Low Energy)

-先把BLE設備放進機車裡,當靠近BLE設備時發出聲音提醒使用者

Put the BLE device in the scooter first, and make a sound to remind the user when approaching the BLE device.
-撰寫Android BLE 程式

Write Android BLE program.

-利用RSSI換成距離,利用距離去判斷是否靠近了機車。

Use RSSI to change distance, use distance to judge whether you are approaching the scooter.



Website of C language teaching

-利用HTML5架構出c語言教學的網頁

 Use HTML5 framework to produce c language teaching web page.


Using Smart Phone accelerometers for activity recognition

-利用手機的三軸加速器收集資料

Use the mobile phone's three-axis accelerator to collect data.
-43個特徵值和k nearest neighbors (KNN)的演算法

43 features ​​and k nearest neighbors (KNN) algorithm

-Android程式,去構成一個可以,辦認坐下、站立走路、騎車、跑步、上樓梯、下樓梯的APP。

Worte android APP that can recognize sitting, standing and walking, cycling, running, upstairs, and downstairs.



Anti theft system based on deep learning

-預計使用pytorch 。(研讀中) 
Expect to use pytorch. (Studying)
-預計會使用論文推薦的CNN model ResNet-50來訓練。(研讀中)
It is expected to use the CNN model ResNet-50 recommended by the paper for training.(Studying)
-將會使用攝影機連結raspberry pi,擷取影像,輸入至test model,去判斷,目前進來的物體,是否為人。(暫時構想)
A camera will be used to connect to the raspberry pi, capture the image, and input it into the test model to determine whether the object currently coming in is a person. (Tentative idea)

Python crawler from Xuite

-在瀏覽相簿時,喜歡的照片必須要手動下載,非常麻煩。

When browsing the album, the favorite photos must be downloaded manually, which is very troublesome

-使用Python的BeautifulSoup抓出html tags將圖片下載下來。

Use Python's BeautifulSoup to grab html tags and download the image



Wireless Sensor Networks Mobile Robots

-碩士論文研究題目:無線感測網路中具能源效率之移動機器人資料收集之研究

Master's thesis research topic: Research on data collection of energy-efficient mobile robots in wireless sensor networks

-研究如何縮短Mobile Robot在物聯網(IoT)裡的無線感測網路裡資料收集路徑的問 題,能達到減少Robot移動上的能量耗損。

Research on how to shorten the data collection path of the Mobile Robot in the wireless sensing network in the Internet of Things (IoT), so as to reduce the energy consumption of Robot movement

-使用Java Swing建構模擬介面。

Use Java Swing to construct a simulation interface

-內部用Java實現最短路徑演算法。

Implement the shortest path algorithm in Java


Search-by-image

-Use tensorflow to train the model with CIFAR-10 wrote by Python.
-目前準確率能到79%,未來會試不同的CNN models,提高準確率。
The current accuracy rate can reach 79%. In the future, different CNN models will be tried to improve the accuracy rate.
-把tensorflow的model轉成Open Neural Network Exchange (ONNX )的格式。
Convert tensorflow model to Open Neural Network Exchange (ONNX) format.
-利用C#撰寫UI,讀取ONNX的model,辨識完圖片後,會依圖片的種類去找相似圖片的資料夾。
Use C# to build UI, read ONNX model to recognize the images, it will find the folder of similar one according to the type of images.