請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/22084
標題: | 電腦視覺與機器學習之穿搭風格判辨-社群媒體來源應用初探 Style classification through Computer Vision and Machine Learning –Originate in social media data base application |
作者: | Ting-Yun Zang 臧廷運 |
指導教授: | 雷欽隆(Chin-Laung Lei) |
關鍵字: | 時尚穿搭,CNN,電腦視覺,機器學習,穿搭風格分類,大數據, fashion,fashion clothing,CNN,computer vision,machine learning,fashion style classification,big data, |
出版年 : | 2018 |
學位: | 碩士 |
摘要: | 隨著科技與經濟蓬勃發展,人們不再只追求基本的物質享受,對於時尚更是日漸 講究。因著社會大眾對於時尚的重視,促使時尚服裝產業朝向更多元的觸角伸 展,結合科技與時尚生活,以及即時分享的特性,網路社群媒體儼然扮演了重要 的角色,人們透過社群媒體貼文分享自己的時尚穿搭逐漸成為常態,形成日常議 題及互相模仿學習,並為時尚產業創造了無限商機。而這些大量的穿搭資訊,憑 藉大數據時代的優勢,使我們得以利用足量的資料進行機器學習,而其中的應用 不乏時尚潮流掌握,時尚穿搭推薦,穿搭風格分類等。
現今在電腦視覺領域的研究日益完整而全面,其中 CNN 被視為一種效果良好的機 器學習方法,不論是物體偵測、影像辨識甚至醫療應用上都佔有重要的地位。然 而在穿搭方面的文獻及研究仍十分有限,相關論文也尚未足量,存在著一定發展 空間。因此本論文便從 CNN 開始針對穿搭風格此一領域進行研究分析,探討 CNN 是否能有效的針對現今流行的穿搭風格進行分類,並在此基礎之上做更進一步分 析與探討延伸。 Due to the dramatic evolution of economic, the fashion industry has developed intensely. People not only pursue for basic needs in their life, but also focus more on fashion outlooks and are willing to spend more money on stylish clothes. Moreover, the social medias haved become more and more popular, people like to share their daily life with others through their posts. For sure, daily outfit is a part of it. Thanks for the social media users, we can learn and imitate a lot of fashionable outfit .Through abundant of wearing-related information and the advantage of big data, it provides enough data for machine learning. Appliances includes grasping the trend of fashion, fashion clothing recommendation, fashion style classification, etc. Now, the research on computer vision has become very comprehensive, and CNN has been proved to be one of the very effective way in machine learning. Appliance such as object detection, video recognition and medical treatment, all of them have important role in their region. However, there isn’t enough paper researching on the region of fashion clothing, which still has much room to improve. This paper focus on the analysis the fashion clothing through CNN, and discuss whether the CNN can effectively classify nowadays fashion styles, and do further research on it. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/22084 |
DOI: | 10.6342/NTU201802612 |
全文授權: | 未授權 |
顯示於系所單位: | 電機工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-107-1.pdf 目前未授權公開取用 | 1.83 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。