装配过程监测图像数据集

数据集内容描述：装配车间装配体三维模型、图片及标注。 数据来源：试验数据。 采集方式：装配体装配步骤严格按照车间装配过程执行，确保每一主要步骤都可采集到相关装配体图像，之后由专人核对检查装配过程及采集图像信息。 加工方式：合成装配体三维模型严格按照装配体主要零部件制作，之后由数据集合成算法按照装配步骤生成装配图像，采集分辨率为256*256；物理装配体图像由Kinect 2.0 采集，分辨率为256*256；图像采集后进行预处理工作。最后，将两类数据集录入数据库并由专人审查核验。 设备情况：采集设备：摄像机（微软 Kinect 2.0），计算机设备（联想Y9000P），服务器（联众集群）。 数据量：75.6BM。Dataset Content Description: 3D models, images and annotations of assemblies in the assembly workshop. Data Source: Experimental data. Collection Method: The assembly steps strictly follow the actual workshop assembly process, ensuring that relevant images of the assembly can be collected for each major step. Subsequently, dedicated personnel will verify and inspect both the assembly process and the collected image information. Processing Method: The synthetic 3D assembly models are strictly manufactured based on the main components of the assembly. Then, the dataset synthesis algorithm generates assembly images according to the assembly steps with a resolution of 256×256. Physical assembly images are collected via Kinect 2.0, also with a resolution of 256×256. Preprocessing is conducted after image collection. Finally, the two types of datasets are imported into the database and reviewed and verified by dedicated personnel. Equipment Details: Collection equipment includes a camera (Microsoft Kinect 2.0), a computer device (Lenovo Y9000P), and a server (Lianzhong Cluster). Data Volume: 75.6 BM.

搜集汇总

背景与挑战

背景概述

该数据集包含装配车间装配体的三维模型、图片及标注，数据来源于试验数据，通过严格遵循装配步骤采集图像，并经过预处理和核验。数据集由合成装配体三维模型和物理装配体图像组成，采集分辨率为256*256，总数据量为75.8MB，包含8997个文件，格式包括csv、png和docx。