基於全身性影像的人物性別與年齡辨識

Abstract

本論文在CRP資料庫集上實作性別與年齡的辨識，研究對象主要是行進中的全身性行人，對行人出現於frame中的所在區域進行分群後，將一連串行人序列在各區域進行辨識後進行整合，得出最後的性別與年齡辨識結果。

演算法方面使用到的特徵為LBP、HOG特徵兩大類，並使用SVM做為辨識的分類器，最後我們也在性別與年齡上，各自以一套系統化的評估方式，來對本論文所做的全部實驗，分析結果的評估好壞。

Gender and age recognition of people are implemented on the CRP database in this paper. The main object of study is the moving full body of pedestrian. After the pedestrians are located in the frame by each region, the sequence of pedestrians are recognized in each region integration to arrive at the final gender and age recognition results. The features used in the algorithm are two major categories of LBP and HOG features, and SVM is used as a classifier for recognition. Finally, we also make a systemic evaluation to evaluate all experiments in this paper, to analyze the results of gender and age recognition which are good or bad.