- 博客(7)
- 收藏
- 关注
原创 StNet: Local and Global Spatial-Temporal Modeling for Action Recognition
论文:https://arxiv.org/abs/1811.01549 代码:https://github.com/PaddlePaddle/models/tree/develop/PaddleCV/PaddleVideo 视频分类算法,AAAI2019的文章 我们知道,视频识别主流的算法方案大致有三,如下所示: 1、对视频进行时序抽样并分别送入CNN提取各自空间信息,得出特征序列后...
2019-12-22 12:45:34
873
原创 跨镜追踪(行人重识别Person Re-identification)读书笔记(四)
《A Coarse-to-fine Pyramidal Model for Person Re-identification via Multi-Loss Dynamic Training》 论文地址:https://arxiv.org/pdf/1810.12193.pdf 腾讯的一篇文章,提出一种多尺寸金字塔模型。如下所示,输入图片经过backbone后得到C × H × W的feature...
2019-01-27 08:55:22
1935
原创 跨镜追踪(行人重识别Person Re-identification)读书笔记(三)
《Parameter-Free Spatial Attention Network for Person Re-Identification》 论文地址:https://arxiv.org/pdf/1811.12150.pdf GitHub代码:https://github.com/HRanWang/Spatial-Attention 文章指出,全局平均池化Global average po...
2019-01-22 18:19:38
1588
1
原创 跨镜追踪(行人重识别Person Re-identification)读书笔记(二)
《Batch Feature Erasing for Person Re-identification and Beyond》 论文地址:https://arxiv.org/pdf/1811.07130.pdf GitHub代码:https://github.com/daizuozhuo/batch-feature-erasing-network 阿里的一篇文章,思路挺新颖的。提出一种新的训...
2019-01-20 11:07:30
1939
原创 跨镜追踪(行人重识别Person Re-identification)读书笔记(一)
《AlignedReID: Surpassing Human-Level Performance in Person Re-Identification》 论文地址:https://arxiv.org/pdf/1711.08184v1.pdf GitHub复现代码:https://github.com/huanghoujing/AlignedReID-Re-Production-Pytorch...
2018-12-14 22:53:31
2989
原创 C++应用中调用YOLOv3(darknet)进行目标检测
YOLOv3论文:https://pjreddie.com/media/files/papers/YOLOv3.pdf 官网和代码:https://pjreddie.com/darknet/ yolo属于one-stage(检测一步到位),兼顾准确率和速度,特别是最近的v3版本提高了小目标的检测率,是移动端目标检测的热门算法。关于YOLO原理的介绍网上有很多资料请自行百度,本文主要介绍如何在自...
2018-12-08 12:25:20
20295
131
原创 人群密度统计网络思路笔记
1、CNN-based Cascaded Multi-task Learning of High-level Prior and Density Estimation for Crowd Counting Two sub tasks: crowd count classification and density map estima...
2018-12-02 19:23:49
5201
1
空空如也
空空如也
TA创建的收藏夹 TA关注的收藏夹
TA关注的人