Hopkins155

Hopkins155数据集由包含两到三个动作的156视频序列组成。每个视频序列的运动对应一个低维子空间。每个视频序列具有来自两个或三个运动提取的数据向量的39-550个视频序列。 序列大致可分为三类: 棋盘序列: 此类别包括在受控条件下用手持相机拍摄的室内场景的104序列。对象上的棋盘图案用于确保大量跟踪点。 交通序列: 此类别由移动手持摄像机捕获的38个室外交通场景序列组成。 其他 (铰接式/非刚性序列): 此类别包含13个序列，显示受关节约束的运动，头部和面部运动，人类行走等。The Hopkins155 dataset comprises 156 video sequences, each containing two or three distinct motions. Each motion in a video sequence corresponds to a low-dimensional subspace. Each video sequence contains 39 to 550 data vectors extracted from its two or three motions. The sequences can be roughly categorized into three classes: Chessboard sequences: This category includes 104 sequences of indoor scenes captured by a hand-held camera under controlled conditions. The chessboard pattern on objects is used to ensure a large number of tracked points. Traffic sequences: This category consists of 38 sequences of outdoor traffic scenes captured by a moving hand-held camera. Others (articulated/non-rigid sequences): This category contains 13 sequences showing joint-constrained motions, such as head and facial movements, human walking, and so on.

搜集汇总

背景与挑战

背景概述

Hopkins155数据集包含156个视频序列，每个序列对应一个低维子空间，数据向量数量在39到550之间。序列分为棋盘序列（104个室内场景）、交通序列（38个室外场景）和其他序列（13个关节/非刚性运动），由约翰霍普金斯大学于2007年发布。