本文選題：場(chǎng)景自適應(yīng)　切入點(diǎn)：行人檢測(cè)　出處：《東南大學(xué)學(xué)報(bào)(自然科學(xué)版)》2017年04期

【摘要】：針對(duì)現(xiàn)有基于機(jī)器學(xué)習(xí)的行人檢測(cè)算法存在當(dāng)訓(xùn)練樣本和目標(biāo)場(chǎng)景樣本分布不匹配時(shí)檢測(cè)效果顯著下降的缺陷,提出一種基于深度模型的場(chǎng)景自適應(yīng)行人檢測(cè)算法.首先,受Bagging機(jī)制啟發(fā),以相對(duì)獨(dú)立源數(shù)據(jù)集構(gòu)建多個(gè)分類器,再通過(guò)投票實(shí)現(xiàn)帶置信度度量的樣本自動(dòng)選取;其次,利用DCNN深度結(jié)構(gòu)的特征自動(dòng)抽取能力,加入一個(gè)自編碼器對(duì)源-目標(biāo)場(chǎng)景下特征相似度進(jìn)行度量,提出了一種基于深度模型的場(chǎng)景自適應(yīng)分類器模型并設(shè)計(jì)了訓(xùn)練方法.在KITTI數(shù)據(jù)庫(kù)的測(cè)試結(jié)果表明,所提算法較現(xiàn)有非場(chǎng)景自適應(yīng)行人檢測(cè)算法具有較大的優(yōu)越性;與已有的場(chǎng)景自適應(yīng)學(xué)習(xí)算法相比較,該算法在檢測(cè)率上平均提升約4%.
[Abstract]:In view of the shortcomings of the existing pedestrian detection algorithms based on machine learning, when the distribution of training samples and target scene samples mismatch, a scene adaptive pedestrian detection algorithm based on depth model is proposed. Inspired by Bagging mechanism, several classifiers are constructed from relative independent source data sets, and then automatic sample selection with confidence measure is realized by voting. Secondly, the feature extraction ability of DCNN depth structure is used. Adding a self-encoder to measure feature similarity in source-target scenarios, a scene adaptive classifier model based on depth model is proposed and a training method is designed. The test results in KITTI database show that, Compared with the existing scene adaptive learning algorithm, the proposed algorithm has more advantages than the existing non-scene adaptive pedestrian detection algorithm, and the average detection rate of the proposed algorithm is increased by about 4% compared with the existing scene adaptive learning algorithm.
【作者單位】： 江蘇大學(xué)汽車工程研究院;江蘇大學(xué)汽車與交通工程學(xué)院;
【基金】：國(guó)家自然科學(xué)基金資助項(xiàng)目(U1564201,61403172,61601203) 中國(guó)博士后基金資助項(xiàng)目(2014M561592,2015T80511) 江蘇省重點(diǎn)研發(fā)計(jì)劃資助項(xiàng)目(BE2016149) 江蘇省自然科學(xué)基金資助項(xiàng)目(BK20140555) 江蘇省六大人才高峰資助項(xiàng)目(2014-DZXX-040,2015-JXQC-012)
【分類號(hào)】：TP18;TP391.41

【相似文獻(xiàn)】

相關(guān)碩士學(xué)位論文 前1條

1 陳浩;基于中層語(yǔ)義特征表達(dá)的物體檢測(cè)方法研究[D];北京工業(yè)大學(xué);2016年

，

本文編號(hào)：1658082