Wood Defect Detection Dataset
收藏资源简介:
Wood Defect Detection数据集是一个为木材制造工业提供的语义分割和目标检测数据集,包含20276张带有语义分割和边界框标注的图像,涉及10种不同的木材缺陷类别。该数据集用于训练扩散模型和分割模型,以生成高质量的工业数据集样本。该数据集有助于降低工业数据标注成本,提高标注质量,并推动计算机视觉模型在现实世界环境中的可靠性和效率。
The Wood Defect Detection Dataset is a semantic segmentation and object detection dataset developed for the wood manufacturing industry. It contains 20,276 images annotated with both semantic segmentation masks and bounding boxes, encompassing 10 distinct wood defect categories. This dataset is designed for training diffusion models and segmentation models to generate high-quality industrial dataset samples. It helps reduce the cost of industrial data annotation, improve annotation quality, and enhance the reliability and efficiency of computer vision models in real-world industrial scenarios.
数据集概述
基本信息
- 数据集名称: Bounding Box-Guided Diffusion for Industrial Image Synthesis
- 论文标题: Bounding Box-Guided Diffusion for Synthesizing Industrial Images and Segmentation Map
- 论文链接: https://arxiv.org/abs/2505.03623
- 会议信息: Synthetic Data for Computer Vision Workshop - CVPR 2025
项目简介
该项目提出了一种基于边界框引导的扩散生成框架,用于合成高质量的工业图像及对应的分割图。该方法支持精确定位、多部件控制和掩模生成,旨在为缺陷检测和分割等下游任务提供数据集支持。
主要特性
- 基于边界框的条件图像合成
- 联合生成分割图
- 支持多对象类型和类别
项目状态
- 开发状态: 项目正在积极开发中,功能、代码结构和文档可能会发生变化。
引用信息
如果使用此代码,请引用以下论文:
@article{simoni2025syth, title={Bounding Box-Guided Diffusion for Synthesizing Industrial Images and Segmentation Maps}, author={Simoni, Alessandro and Pelosin, Francesco}, booktitle={Computer Vision and Pattern Recognition Workshops 2025 (CVPR 2025)}, year={2025} }




