- 學位論文
基於YOLO演算法即時偵測移動車輛
The object detection of moving ground vehicles using YOLO algorithm on UAV
摘要
電腦視覺的定義之一，首先想到的就是圖像分類。沒錯，圖像分類是物件偵測最基本的任務之一，但是在圖像分類的基礎上，還有更複雜和有意義的任務，如物件偵測，物件定位，圖像分割等; 其中物件偵測是一件比較實際的且具有挑戰性的電腦視覺任務，其可以看成圖像分類與定位的結合，給定一張圖片，物件偵測系統要能夠識別出圖片的目標並給出其位置，由於圖片中目標數是不定的，且要給出目標的精確位置，物件偵測相比分類任務更複雜。物件偵測的一個實際應用場景就是自動駕駛和無人機(UAV)，如果能夠在測試機台裝載一個有效的物件偵測系統，如同人類有眼睛，可以快速地檢測出行人與車輛，從而作出適時決策。本研究的目的是開發一個物件偵測系統並裝置在UAV，即時偵測牆壁上特殊的符號及地面上車輛，並於遠端的系統上操作與控制。本研究運用YOLO演算法進行一系列的實作。
並列摘要
When people talk about the definition of Computer Vision, the first thing that comes to mind is the image classification. Previous researches illustrated that image classification is one of the most basic tasks of computer vision. However, based on the basis of image classification, there are more complicated and interesting tasks, such as: object detection, object location, image segmentation...etc in computer vision. The object detection is a practical and challenging task, which can be regarded as a combination of image classification and object location, given a complicated target detection system. To identify the selected object from various targets in the picture (target detection system), and to give the precise location of the target demonstrated the target detection is more complicated than the classification task. A practical application scenario of target detection is autonomous cars and Unmanned Aerial Vehicle (UAV). The aim of this research is to develop an object detection system on a UAV. The developed system is capable to real-time capture the wanted target on the wall and moving target on the ground using remote control system. A series of comprehensive tests has been conducted based on the YOLO algorithm in this paper.