当前位置:网站首页>视频异常检测数据集 (ShanghaiTech)
视频异常检测数据集 (ShanghaiTech)
2022-06-23 16:49:00 【因吉】
1 引入
源地址:https://svip-lab.github.io/dataset/campus_dataset.html
论文地址:https://openaccess.thecvf.com/content_cvpr_2018/papers/Liu_Future_Frame_Prediction_CVPR_2018_paper.pdf
数据集及代码:https://github.com/StevenLiuWen/ano_pred_cvpr2018
2 概述
异常检测模型训练的一个直接目标是可以直接应用于多视角多场景中。然而,几乎所有已有数据集都只包含使用一台固定角度相机拍摄的视频,且缺乏场景和视角的多样性。为了增加场景多样性,构建了一个新的异常检测数据集ShanghaiTech。此外,在该数据集中引入了由突然运动引起的异常,例如追逐和斗殴。这些异常是现有数据集不具备的,使得所构建的数据集更适合真实场景。
异常检测数据集总结如下:
1)CUHK Avenue:包含16个训练视频和21个测试视频,共47个异常事件,包含投掷物体、闲逛和跑步。人的大小会因为相机的位置和角度而改变;
2)Pedestrian 1 (Ped1):包含34个训练视频和36个测试视频,其中包含40个不规则事件。所有这些异常案例都与自行车、汽车等交通工具有关。
3)Pedestrian 2 (Ped2):包含16个训练视频和12个测试视频,包含12个异常事件。Ped2 的异常定义与 Ped1 相同。
4)Subway:有两类,即入口和出口。不寻常的事件包括走错方向和游荡。该数据集是在室内环境中记录的,而以上数据是在室外环境中记录。
5)ShanghaiTech:包含13个场景,具有复杂的光照条件和摄像机角度。包含130 个异常事件和超过270000个训练帧。此外,异常事件的像素级标注被给出。
3 Bib
@inproceedings{
Liu:2018:65366545,
author = {
Wen Liu and Wei Xin Luo and Dong Ze Lian and Sheng Hua Gao},
title = {
Future frame prediction for anomaly detection--a new baseline},
booktitle = {
{
IEEE} conference on Computer Vision and Pattern Recognition},
pages = {
6536--6545},
year = {
2018}
}
边栏推荐
- 手机开户流程是什么?现在网上开户安全么?
- C#与数据库连接
- History of storage technology: from tape to hardware liquefaction
- Bypass rights
- Revil - blackmail Virus Emergency Response
- torch学习(一):环境配置
- Li Kou daily question - day 25 -495 Timo attack
- Nodejs implements multi process
- January 5, 2022: there are four kinds of rhythms: AABB, ABAB and ABB
- First use of kubernetes cronjob
猜你喜欢

Query the size of each table in the database

数据库 实验二 查询
![[network communication -- webrtc] source code analysis of webrtc -- bandwidth estimation at the receiving end](/img/b0/97dbf3d07a4ed86d6650a58a97a5fc.png)
[network communication -- webrtc] source code analysis of webrtc -- bandwidth estimation at the receiving end

MySQL transaction and its characteristics and locking mechanism

Crmeb second open SMS function tutorial

hands-on-data-analysis 第二单元 第四节数据可视化

Self supervised learning (SSL)

torch学习(一):环境配置

12 initialization of beautifulsoup class
![[go] calling Alipay to scan code for payment in a sandbox environment](/img/d4/c6d72a697bc08f69f11121a15109b3.png)
[go] calling Alipay to scan code for payment in a sandbox environment
随机推荐
PostgreSQL series articles -- the world's most advanced open source relational database
What is the mobile account opening process? Is it safe to open an account online now?
《AN4190应用笔记 天线选择指南》——天线理论2
Troubleshooting of datanode entering stale status
[mae]masked autoencoders mask self encoder
How to quickly obtain and analyze the housing price in your city?
Self supervised learning (SSL)
Hands on data analysis unit 2 section 4 data visualization
手机开户一般哪个证券公司好?在线开户安全么?
Li Kou daily question - day 25 -495 Timo attack
Database Experiment 2 query
如何通过线上股票开户?在线开户安全么?
MySQL - reasons for using repeatable read
Revil - blackmail Virus Emergency Response
Best practices cloud development cloudbase content audit capability
MySQL事务提交流程
New function! Qianfan magic pen apaas December capability monthly report
[30. concatenate substrings of all words]
What does the timestamp 90K mean?
This time, thoroughly understand the SparseArray implementation principle