T-LESS
收藏资源简介:
T-LESS数据集是由捷克技术大学机器感知中心创建的,专注于无纹理刚性物体的6D姿态估计。该数据集包含30个工业相关对象,这些对象没有显著的纹理和颜色或反射特性,且在形状和/或大小上表现出对称性和相互相似性。数据集包括约39,000个训练图像和10,000个测试图像,这些图像由三个同步传感器捕获,包括结构光和飞行时间RGB-D传感器以及高分辨率RGB相机。每个对象提供两种类型的3D模型:手动创建的CAD模型和半自动重建的模型。训练图像显示单个对象与黑色背景,而测试图像来自20个具有不同复杂度的测试场景,从简单的场景到具有多个对象实例和大量杂乱和遮挡的非常具有挑战性的场景。T-LESS数据集旨在评估6D对象姿态估计问题的各种变体,以及其他相关问题,如2D对象检测和对象分割。
The T-LESS dataset was created by the Center for Machine Perception at the Czech Technical University, focusing on 6D pose estimation of texture-less rigid objects. It contains 30 industrial-related objects that have no prominent texture, color or reflection properties, and exhibit symmetry and mutual similarity in shape and/or size. The dataset includes approximately 39,000 training images and 10,000 test images captured by three synchronized sensors: structured light and time-of-flight RGB-D sensors, as well as a high-resolution RGB camera. Two types of 3D models are provided for each object: manually created CAD models and semi-automatically reconstructed models. Training images show single objects against a black background, while test images are sourced from 20 test scenes with varying complexity, ranging from simple scenarios to highly challenging ones with multiple object instances, extensive clutter and occlusion. The T-LESS dataset is designed to evaluate various variants of the 6D object pose estimation task, as well as other related problems such as 2D object detection and object segmentation.

- 1T-LESS: An RGB-D Dataset for 6D Pose Estimation of Texture-less Objects捷克技术大学机器感知中心 · 2017年



