WPI

WPI 数据集是一个对交通信号灯、行人和车道数据进行视频记录并分析的数据集。
数据集包括：
WPI 交通信号灯数据集：（Frames_GT_wHolder：带有标签的全部帧。每个文件夹都是一种带有持有者的交通灯类型，并且只有该类型被标记 ( Ground truth 包含交通灯持有者)。一个框架包含多种类型的交通灯可能出现在不同的文件夹中；Frames_GT_Lights：带有标签的全部帧。这些序列被标记为所有类型的单个交通灯。序列 8、9、10 没有红绿灯）WPI 行人道数据集：19 个测试序列，29 个训练序列WPI 车道保持数据集：提供相机参数、图像、速度和方向盘角度测试数据是在美国马萨诸塞州伍斯特市的冬季收集的。

The WPI dataset is a dataset for video recording and analysis of traffic light, pedestrian and lane data. The dataset includes the following components:
1. WPI Traffic Light Dataset:
- Frames_GT_wHolder: All labeled frames. Each folder corresponds to a type of traffic light with its holder, and only that specific type is annotated (the Ground Truth includes the traffic light holder). A single frame may contain multiple types of traffic lights, which will be categorized into different folders;
- Frames_GT_Lights: All labeled frames. These sequences are annotated with individual traffic lights of all types. Sequences 8, 9 and 10 have no traffic lights.
2. WPI Pedestrian Dataset: 19 test sequences and 29 training sequences.
3. WPI Lane Keeping Dataset: Provides camera parameters, images, speed and steering angle data.
This dataset was collected during winter in Worcester, Massachusetts, USA.

搜集汇总

背景与挑战

背景概述

WPI数据集是一个由伍斯特理工学院发布的交通分析数据集，包含交通信号灯、行人和车道保持三个子集，数据在美国伍斯特市冬季收集，适用于自动驾驶或交通监控研究。该数据集提供了详细的视频帧标签和传感器数据，但部分序列可能缺少交通灯信息。