《電子技術(shù)應(yīng)用》
您所在的位置:首頁 > 人工智能 > 設(shè)計(jì)應(yīng)用 > 全局通道注意力增強(qiáng)的毫米波圖像目標(biāo)檢測
全局通道注意力增強(qiáng)的毫米波圖像目標(biāo)檢測
電子技術(shù)應(yīng)用
蔣甜甜,,葉學(xué)義,李剛,,楊夢豪,,陳華華
杭州電子科技大學(xué) 通信工程學(xué)院,浙江 杭州 310018
摘要: 針對主動毫米波圖像中目標(biāo)與背景紋理區(qū)分度較低導(dǎo)致隱匿目標(biāo)漏檢問題,,并根據(jù)安檢實(shí)時(shí)性要求,,提出一種基于全局通道注意力增強(qiáng)的主動毫米波圖像目標(biāo)檢測方法。該方法以YOLOv5s為載體,,在坐標(biāo)注意力位置方向上引入全局通道注意模塊,,增強(qiáng)對隱匿目標(biāo)全局通道信息的關(guān)注,從而提升在隱匿目標(biāo)與背景紋理區(qū)分度較低時(shí)的檢測能力,;再利用K-means++聚類算法重新生成適合毫米波圖像目標(biāo)檢測的錨框,。實(shí)驗(yàn)結(jié)果表明,無論是陣列圖像數(shù)據(jù)集還是線掃圖像數(shù)據(jù)集,,該方法增強(qiáng)了對隱匿目標(biāo)的特征注意,,提高了召回率,在滿足安檢實(shí)時(shí)性的前提下,,提升了檢測性能,。通過增加少量參數(shù),在陣列圖像數(shù)據(jù)集上,,精度,、召回率和[email protected]達(dá)到了92.0%、90.93%和95.32%;在線掃圖像數(shù)據(jù)集上,,精度,、召回率和[email protected]達(dá)到了94.65%、92.67%和97.73%,。平均單張圖像推理時(shí)間在兩個數(shù)據(jù)集上均達(dá)到1 ms,,滿足實(shí)時(shí)性要求。
中圖分類號:TP391.4 文獻(xiàn)標(biāo)志碼:A DOI: 10.16157/j.issn.0258-7998.234502
中文引用格式: 蔣甜甜,,葉學(xué)義,,李剛,等. 全局通道注意力增強(qiáng)的毫米波圖像目標(biāo)檢測[J]. 電子技術(shù)應(yīng)用,,2024,,50(3):19-25.
英文引用格式: Jiang Tiantian,Ye Xueyi,,Li Gang,,et al. Global channel attention boosted active millimeter wave image object detection[J]. Application of Electronic Technique,2024,,50(3):19-25.
Global channel attention boosted active millimeter wave image object detection
Jiang Tiantian,,Ye Xueyi,Li Gang,,Yang Menghao,,Chen Huahua
School of Communication Engineering, Hangzhou Dianzi University,, Hangzhou 310018,, China
Abstract: Due to the low discrimination between objects and background texture in active millimeter wave images and the need for security in real time, a global channel attention booster-based method for active millimeter wave image object detection is proposed. In order to improve attention to the global channel information of the concealed object and improve detection performance when the concealed object could not be distinguished from the background texture, this method uses YOLOv5s as the carrier and adds global channel attention to the position direction of coordinate attention. And the K-Means ++ clustering method is used to create the anchor box for identifying concealed objects in millimeter wave images. The results demonstrate that both for array image dataset and line sweep image dataset, the detection model enhances the attention of hidden objects feature and improves the detection performance on the basis of meeting the security real-time performance.
Key words : active millimeter wave image detection;global channel attention booster,;K-means++,;attention mechanism

引言

人們對自身安全問題的持續(xù)關(guān)注,,促進(jìn)了在公共場所以及一些重點(diǎn)場所安檢技術(shù)的發(fā)展,。傳統(tǒng)的安檢技術(shù)包括X射線探測器、金屬探測器[1]以及人工檢測[2],,它們在人體健康和隱匿目標(biāo)檢測方面存在一定限制,。由于毫米波輻射是非電離輻射[3],對衣物穿透性強(qiáng)[4],,能夠在不傷害人體情況下進(jìn)行安全檢查,,因而毫米波安檢技術(shù)越發(fā)受到關(guān)注;且隨著成像技術(shù)和計(jì)算機(jī)視覺的飛速發(fā)展,,毫米波安檢技術(shù)逐漸與人工智能方法相結(jié)合,。因?yàn)橹鲃雍撩撞?Active Millimeter Wave, AMMW)[5]圖像質(zhì)量要高于被動毫米波(Passive Millimeter Wave, PMMW)[6],對AMMW圖像(包括陣列和線掃毫米波成像設(shè)備)的隱匿目標(biāo)檢測逐漸成為主流。


本文詳細(xì)內(nèi)容請下載:

http://forexkbc.com/resource/share/2000005910


作者信息:

蔣甜甜,,葉學(xué)義,,李剛,楊夢豪,,陳華華

杭州電子科技大學(xué) 通信工程學(xué)院


雜志訂閱.jpg

此內(nèi)容為AET網(wǎng)站原創(chuàng),,未經(jīng)授權(quán)禁止轉(zhuǎn)載。