室内弓字形轨迹合成数据

本数据集专为开发与提升扫地机器人的智能水平而设计，通过高度模拟其实际工作时的“弓字形”清洁路径，为机器人视觉算法的研究与训练提供全方位的仿真数据支持。数据集以一个真实的室内家装场景为背景，核心特点是完全模拟了扫地机器人的第一视角工作状态。所有数据均沿着一条典型的“弓字形”清洁轨迹生成，相机参数严格设定为：机身半径150mm，镜头高度仅50mm（紧贴地面），视角为60度。这种极低的视角完美复现了扫地机器人在行进中所“看到”的景象，包括家具底部、桌椅腿、墙角、电源线等极具挑战性的狭窄和遮挡环境，极大地增强了数据在真实应用中的价值。
我们选取了一个室内的家装场景，模拟机身半径为150mm扫地机器人进行家庭清洁时的轨迹来生成相机点位进行渲染，渲染的参数设置为，分辨率1920*1080，fov60,相机高度50mm。数据集内包含以下类型的内容：相机位姿（内外参），深度图，coco格式2d图片标注信息， 相机坐标系下的法向图，渲染图，语义图，albedo通道图。

This dataset is specifically designed for developing and improving the intelligent capabilities of sweeping robots. By highly simulating the actual "zigzag" cleaning path during their routine operation, it provides comprehensive simulation data support for the research and training of robot vision algorithms.
The dataset is based on a real indoor home decoration scene, and its core feature is the full simulation of the first-person working perspective of sweeping robots. All data is generated along a typical "zigzag" cleaning trajectory, with the camera parameters strictly configured as follows: body radius 150mm, lens height of only 50mm (extremely close to the ground), and a field of view of 60°. This extremely low shooting angle perfectly reproduces the scenes that the sweeping robot "sees" while navigating, including challenging narrow and occluded environments such as furniture undersides, table and chair legs, room corners, power cords, etc., greatly improving the practical value of the dataset for real-world deployment.
We selected an indoor home decoration scene to generate camera positions for rendering by simulating the cleaning trajectory of a 150mm-body-radius sweeping robot during household cleaning. The rendering parameters are configured as: resolution 1920×1080, FOV 60°, and camera height 50mm.
The dataset contains the following types of data: camera poses (internal and external parameters), depth maps, COCO-format 2D image annotation information, normal maps in the camera coordinate system, rendered images, semantic maps, and albedo channel maps.

搜集汇总

背景与挑战

背景概述

该数据集的名称为'室内弓字形轨迹合成数据'，但提供的详情内容主要涉及温岭市人群的肠镜和胃镜检测数据，这些数据由台州市肿瘤医院申请，用于结直肠癌筛查与早期诊断。内容描述了多个特定部位（如横结肠、回肠、胃窦等）的检测数据集，每个数据集都通过量化指标（如病变部位、病理诊断编码、腺瘤性息肉绒毛状结构比例和最大径）进行评分，并基于AHP层次法将风险分为高、中、低三个等级，以支持临床决策和标准化报告生成。