视频处理_Caltech Pedestrian Dataset(加利福尼亚理工学院行人数据库)
- 1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
- 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
- 3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。
Caltech Pedestrian Dataset(加利福尼亚理工学院行人
数据库)
数据摘要：
The Caltech Pedestrian Dataset consists of approximately 10 hours of 640x480 30Hz video taken from a vehicle driving through regular traffic in an urban environment. About 250,000 frames (in 137 approximately minute long segments) with a total of 350,000 bounding boxes and 2300 unique pedestrians were annotated. The annotation includes temporal correspondence between bounding boxes and detailed occlusion labels. More information can be found in our CVPR09 benchmarking paper.
中文关键词：
行人,检测,时序对应,包围盒,遮挡标记,
英文关键词：
Pedestrian,detection,temporal correspondence,bounding
boxes,occlusion labels,
数据格式：
VIDEO
数据用途：
To detection pedestrian from video
数据详细介绍：
Caltech Pedestrian Dataset
Description
The Caltech Pedestrian Dataset consists of approximately 10 hours of 640x480 30Hz video taken from a vehicle driving through regular traffic in an urban environment. About 250,000 frames (in 137 approximately minute long segments) with a total of 350,000 bounding boxes and 2300 unique pedestrians were annotated. The annotation includes temporal correspondence between bounding boxes and detailed occlusion labels. More information can be found in our CVPR09 benchmarking paper.
Download
Caltech Pedestiran Dataset. The training data (set00-set05) consists of six training sets (~1GB each), each with 6-13 one-minute long seq files, along with all annotation information (see the paper for details). Detection results for all evaluated algorithms are also provided. The testing images (set06-set10), but not the annotations, are now also provided, please see "submitting results" below for information on how to include your trained pedestrian detector in the evaluation.
Seq video format. An seq file is a series of concatenated image frames with a fixed size header. Matlab routines for reading/writing/manipulating seq files can be found in Piotr's Matlab Toolbox (version 2.51 or later recommended).
Matlab evaluation/labeling code (2.2.0). The annotations use a custom "video bounding box" (vbb) file format. The code also contains utilities to view seq files with annotations overlayed, evaluation routines used to generate all the ROC plots in the paper, and also the vbb labeling tool used to create the dataset (see also this somewhat outdated video tutorial).
Additional datasets in standardized format. For convenience we are posting full images/annotations in seq/vbb format as well as detection results for all evaluated algorithms on a number of additional datasets. This facilitates training/testing on these additional datasets and exact reproduction of all ROC curves. Full copyright remains with the original authors, please see the respective website for additional information including how to cite evaluation results on these datasets. INRIA pedestrian dataset (converted), ETH pedestrian dataset (converted), TUD-Brussels pedestrian dataset (converted), Daimler pedestrian dataset (converted).