行人重識別中,很多選擇resnet50作爲backbone,但是會把第四層的stride改爲1,獲得到更大的特徵map,以下是程序。
鏈接: https://github.com/Henuzhaoyli/resnet-50-.
RESNET 修改第四層步長爲1,得到更大的size
發表評論
所有評論
還沒有人評論,想成為第一個評論的人麼? 請在上方評論欄輸入並且點擊發布.
相關文章
1606.End--to-End Comparative Attention Networks for Person Re-identification 論文筆記
xuluohongshang
2020-06-13 20:24:16
Person Re-identification based on Two-Stream Network with Attention and Pose Features 論文總結筆記
若比达法则
2020-07-08 06:30:38
行人重識別(Person Re-ID)【二】:基於深度學習的行人重識別研究綜述
volcano_Lin
2020-07-03 06:07:29
利用memory bank 存儲特徵,進行REID跨域問題的研究
Tchunren
2020-07-02 14:49:17
行人重識別(Person re-identification)概述
BruceZhou95
2020-06-26 04:01:39
Contribution-Based Multi-Stream Feature Distance Fusion Method With K-Distribution Re-Ranking re-id
若比达法则
2020-06-24 20:58:46
簡單行人重識別代碼到88%準確率 鄭哲東 準備工作
asiwxy
2020-06-22 23:32:52
2016 ECCV-Gated Siamese Convolutional Neural Network Architecture for Human Re-ID
_Xiaobo
2020-06-22 01:45:24
行人重識別0-00:DG-Net(ReID)-目錄-史上最新無死角講解
江南才尽,年少无知!
2020-06-17 02:48:41
轉載:GANS的世界2-0:DG-Net(行人重識別ReID)-目錄-史上最新無死角講解
cq99312254
2020-06-16 03:50:10
Scalable Person Re-identification: A Benchmark(可擴張的行人重識別:基準)-1
cq99312254
2020-06-16 03:50:10
GLAD: Global-Local-Alignment Descriptor for Pedestrian Retrieval 論文閱讀筆記
若比达法则
2020-06-16 03:12:42
【KD】、【reid】Distilled Person Re-identification: Towards a More Scalable System
moonuke
2020-06-13 20:26:18