本文主要提出一個以主從式架構之視訊標記系統。此系統可以應用於道路安全監控之物件分
類上。我們根據實際場景之需求不同，發展出適合於大物件及小物件兩種攝影機視角之分類演算
法，演算法中將物件標記為汽車、機車及行人三種特定之物件類別。此外，本主從式系統可透過
網路提供多個使用者連線操作，將標記結果儲存於伺服器資料庫中。由實驗結果驗證此系統可提
供一個彈性且有效率之視訊標記方法。

This paper proposes a client/server-based video indexing system which can be used for the object classification in road surveillance. Based on various requirements of different scenarios, we have developed
an efficient classification algorithm considering the factors of object sizes and camera viewpoints. Three different objects, namely vehicle, motorcycle and pedestrian can be precisely identified. Furthermore,
multi-users are able to control the client/server-based video indexing system and store the object classification results to the server database through the internet. The excellent experimental results show that
the proposed system is flexible and effective for video indexing.