Kitchen Scenes Dataset

Kitchen Scenes Dataset是由乔治梅森大学计算机科学系创建的一个新的多视角RGB-D数据集，专注于厨房场景中的物体实例检测和识别。该数据集包含9个真实的厨房环境视频，每个场景中包含多个物体，这些物体来自BigBird数据集和其他常见物品如碗和咖啡杯。数据集使用高分辨率的Kinect V2 RGB-D传感器收集，通过密集采样视角和3D点云中的边界框注释来增强数据集的复杂性。该数据集主要用于解决服务机器人在现实环境中检测和识别常见家用物体的问题，通过与Washington RGBD Scenes (WRGB-D)数据集比较，展示了其在物体检测和识别方面的挑战性。

Kitchen Scenes Dataset is a novel multi-view RGB-D dataset developed by the Department of Computer Science, George Mason University, focusing on object instance detection and recognition in kitchen environments. This dataset includes 9 videos of real kitchen scenarios, with each scene containing multiple objects sourced from the BigBird dataset and other common household items such as bowls and coffee cups. Collected using high-resolution Kinect V2 RGB-D sensors, the dataset enhances its complexity through dense view sampling and bounding box annotations in 3D point clouds. It is primarily designed to address the challenge of service robots detecting and recognizing common household objects in real-world environments, and demonstrates its challenging performance in object detection and recognition by comparison with the Washington RGBD Scenes (WRGB-D) dataset.

搜集汇总

背景与挑战

背景概述

Kitchen Scenes Dataset是一个多视角RGB-D数据集，由乔治梅森大学创建，包含9个真实厨房环境视频，使用Kinect V2传感器收集并标注3D边界框。该数据集专注于厨房场景中的物体实例检测和识别，旨在帮助服务机器人解决现实环境中的物体识别问题，并通过与其他数据集的比较突显其挑战性。