MTCNN演算法提速應用(ARM測試結果評估) MTCNN演算法提速應用(ARM測試結果評估)
阿新 • • 發佈:2018-11-16
原
,速度30ms!
MTCNN演算法提速應用(ARM測試結果評估)
置頂 2017年11月02日 10:48:05 samylee 閱讀數:11584<span class="tags-box artic-tag-box"> <span class="label">標籤:</span> <a data-track-click="{"mod":"popu_626","con":"mtcnn人臉檢測"}" class="tag-link" href="http://so.csdn.net/so/search/s.do?q=mtcnn人臉檢測&t=blog" target="_blank">mtcnn人臉檢測 </a> <span class="article_info_click">更多</span></span> <div class="tags-box space"> <span class="label">個人分類:</span> <a class="tag-link" href="https://blog.csdn.net/samylee/article/category/6921813" target="_blank">人臉檢測 </a> </div> </div> <div class="operating"> </div> </div> </div> </div> <article> <div id="article_content" class="article_content clearfix csdn-tracking-statistics" data-pid="blog" data-mod="popu_307" data-dsm="post"> <div class="article-copyright"> 版權宣告:本文為博主原創文章,歡迎轉載。 https://blog.csdn.net/samylee/article/details/78421960 </div> <link rel="stylesheet" href="https://csdnimg.cn/release/phoenix/template/css/ck_htmledit_views-f76675cdea.css"> <div class="htmledit_views"> <p>經博主測試,mtcnn原三層網路如果用於工程測試,<span style="color:#ff0000;">誤檢情況嚴重</span>,在fddb上測試結果也是,經常將<span style="color:#ff0000;">手或者耳朵</span>檢測為人臉,這個很頭疼(<span style="color:#ff0000;">因為標註資料!</span>),所以重新訓練顯得尤為重要!</p>
博主的改進方法及如何重新訓練的就不具體介紹了,主要思想就是用卷積取代池化,fddb測試離散ROC88!
注意:某些公開的非官方mtcnn訓練方法有誤!只可參考,不可深入!
PC端測試:(測試軟體:vs2015,測試硬體:i7-4790-4core)
1920x1080視訊,最小檢測人臉為60,速度為22ms!
640x480視訊,最小人臉為25,速度為17ms!
arm端測試:(硬體:香橙派,全志A64晶片,4核64位Cortex-A53,市場價格240元!)
測試:640x480視訊,最小檢測人臉80
測試效果如下(這裡對比了Shiqi YU的人臉檢測):
演算法 | 測試影象尺寸 | 測試最小人臉尺寸 | 演算法耗時(ms) |
ShiqiYU-facedetect_frontal | 2064x1078 | 40 | 95 |
ShiqiYU-facedetect_frontal_surveillance | 2064x1078 | 40 | 125 |
ShiqiYU-facedetect_multiview | 2064x1078 | 40 | 215 |
ShiqiYU-facedetect_multiview_reinforce | 2064x1078 | 40 | 380 |
OURS | 2064x1078 | 40 | 83 |
ShiqiYU-facedetect_fronta
ShiqiYU-facedetect_frontal_surveillancel
ShiqiYU-facedetect_multiview
ShiqiYU-facedetect_multiview_reinforce
ours
任何問題請加唯一QQ2258205918(名稱samylee)!