摘要

我们提供了一个新的包含九个办公场景的多视角实例级数据集，在不同的物体遮挡和光照强度下对常见的实例物体进行了多视角的拍摄。对场景视频抽帧（抽帧代码见脚本附件）得到最终的图像数据库共1721张，人工标注矩形框生成标注文件（标注工具如LabelImg等）。

Abstract

We provide a new multi-view instance-level dataset of nine office scenes. Take multi-view shots of common instance objects under different object occlusions and light intensities. Extract frames (code in attachment of scripts) from scene videos to get a total of 1721 images, and objects in the scenes are annotated with bounding boxes (tools such as LabelImg, etc.).

内容

• 9个分辨率为1920*1080的办公场景序列

• 每个场景包含2-7个实例物体，数据库总共7个实例物体

• 图片中所有实例物体均进行了矩形框标注

• 拍摄设备：Canon EOS 80D, 18-135mmIS USM镜头，帧数为30fps

有关更多详细信息，请参阅数据链接中包含的README文件。

Contents

• 9 RGB-D kitchen video sequences with 1920x1080 resolution.

• 2-7 object instances per scene.7 object instances in total across scenes.

• Bounding box annotations for all objects in the RGB-D frames.

• Equipment: Canon EOS 80D, 18-135mmIS USM. Video at 30fps.

For more details please see the README files that are included in the data links.

数据（Data）

办公场景的视频序列及标注文件（Video sequences of the office scenes and annotation )

 README.zip

场景1（Scene 1）： bhid_scene_1.zip

场景2（Scene 2）： bhid_scene_2.zip

场景3（Scene 3）： bhid_scene_3.zip

场景4（Scene 4）： bhid_scene_4.zip

场景5（Scene 5）： bhid_scene_5.zip

场景6（Scene 6）： bhid_scene_6.zip

场景7（Scene 7）： bhid_scene_7.zip

场景8（Scene 8）： bhid_scene_8.zip

场景9（Scene 9）： bhid_scene_9.zip

脚本

将场景视频抽帧得到图片（python）:  extract_frame.zip

Scripts

Transform scene videos to image(python):extract_frame.zip