本文選題：深度神經(jīng)網(wǎng)絡(luò)　切入點(diǎn)：行人再識別　出處：《西南交通大學(xué)》2017年碩士論文　論文類型：學(xué)位論文

【摘要】：行人再識別(Person re-identification)技術(shù)是判斷在不同監(jiān)控?cái)z像頭下出現(xiàn)的行人圖像是否屬于同一行人的技術(shù)。面對海量增長的監(jiān)控視頻,利用計(jì)算機(jī)對監(jiān)控視頻中的行人進(jìn)行再識別的需求應(yīng)運(yùn)而生。然而現(xiàn)存的行人再識別算法主要是在已裁剪好的行人圖片中匹配查找集和候選集,這是不切實(shí)際的,行人的框架在現(xiàn)實(shí)考慮中不可能直接給定,目標(biāo)行人需要在整張圖片中被鎖定。目前,深度學(xué)習(xí)在圖像識別、語音識別、自然語言處理等多個領(lǐng)域取得了優(yōu)異的效果。相比于傳統(tǒng)人工提取特征的方法,深度神經(jīng)網(wǎng)絡(luò)通過從數(shù)據(jù)中自動學(xué)習(xí)到更能表征圖像的特征并進(jìn)行分類,更具實(shí)際意義。將深度學(xué)習(xí)應(yīng)用到行人再識別上已經(jīng)成為當(dāng)前的研究熱點(diǎn),但是由于目前行人再識別中如圖像分辨率低、遮擋、光照變化等問題使其離實(shí)際應(yīng)用還有很長的距離。本文總結(jié)了目前一些行人檢測及再識別的常用特征、算法以及深度神經(jīng)網(wǎng)絡(luò)結(jié)構(gòu),并進(jìn)行深入研究和分析。設(shè)計(jì)了一種針對端到端行人再識別的預(yù)訓(xùn)練網(wǎng)絡(luò)模型,該模型結(jié)合了驗(yàn)證和分類兩種網(wǎng)絡(luò)結(jié)構(gòu),并利用空間池化操作對不同尺度的輸入圖片進(jìn)行特征歸一化。在此基礎(chǔ)上用性能良好的ResNet-50網(wǎng)絡(luò)結(jié)構(gòu)對端到端的行人再識別網(wǎng)絡(luò)結(jié)構(gòu)進(jìn)行改進(jìn)。之后在caffe深度學(xué)習(xí)框架上訓(xùn)練改進(jìn)的模型并進(jìn)行多組實(shí)驗(yàn),包括預(yù)訓(xùn)練模型的有效性、不同特征維度對網(wǎng)絡(luò)模型效果的影響、在不同大小的候選集、低分辨率和遮擋子集下的性能分析,以及與當(dāng)前比較先進(jìn)的算法進(jìn)行對比。實(shí)驗(yàn)結(jié)果證明了本文方法訓(xùn)練出來的模型能夠?qū)W習(xí)到具有較高魯棒性的特征,大幅度提高了行人再識別的識別率。
[Abstract]:Pedestrian re-identification is a technique to determine whether a pedestrian image under different surveillance cameras belongs to the same pedestrian. The need for rerecognition of pedestrians in surveillance videos arises with the help of computers. However, it is unrealistic for existing pedestrian rerecognition algorithms to match and find sets and candidate sets in a cut pedestrian image. The pedestrian frame cannot be given directly in practical considerations, and the target pedestrian needs to be locked in the entire picture. At present, the depth of learning is in image recognition, speech recognition, Natural language processing and other fields have achieved excellent results. Compared with the traditional methods of artificial feature extraction, depth neural networks can automatically learn from the data to represent the features of images and classify them. The application of depth learning to pedestrian rerecognition has become a hot research topic. However, due to the low image resolution and occlusion in pedestrian rerecognition, This paper summarizes some common features, algorithms and depth neural network structure of pedestrian detection and re-recognition. A pre-training network model for end-to-end pedestrian rerecognition is designed, which combines verification and classification of network structure. The spatial pool operation is used to normalize the input images of different scales. On the basis of this, the end-to-end pedestrian recognition network structure is improved with the good performance of ResNet-50 network structure, and then the caffe depth learning is carried out. Training the improved model on the framework and conducting multiple groups of experiments, Including the effectiveness of the pre-training model, the influence of different feature dimensions on the effectiveness of the network model, and the performance analysis under different sizes of candidate sets, low resolution and occlusion subsets. The experimental results show that the model trained by this method can learn the characteristics of high robustness and greatly improve the recognition rate of pedestrian recognition.
【學(xué)位授予單位】：西南交通大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：TP391.41

【參考文獻(xiàn)】

相關(guān)期刊論文 前3條

1 胡聰叢;胡桓;;深度神經(jīng)網(wǎng)絡(luò)的發(fā)展現(xiàn)狀[J];電子技術(shù)與軟件工程;2017年04期

2 仇春春;楊星紅;程海粟;郭晶晶;;基于特征表示的行人再識別技術(shù)綜述[J];信息技術(shù);2016年07期

3 俞婧;仇春春;王恬;許金鑫;;基于距離匹配的行人再識別技術(shù)綜述[J];微處理機(jī);2016年03期

，

本文編號：1641914