Visual Wake Words
收藏arXiv2019-06-13 更新2024-06-21 收录
下载链接：
https://github.com/tensorflow/models/blob/master/research/slim/datasets/build_visual_wake_words_data.py
下载链接
链接失效反馈官方服务：
资源简介：
Visual Wake Words数据集是由谷歌研究团队创建，旨在为微控制器视觉应用提供一个基准数据集，特别关注于识别图像中是否存在人物。该数据集包含115,000张训练和验证图像，来源于公开的COCO数据集，通过重新标记以区分人物和非人物类别。创建过程涉及使用特定脚本过滤和标记COCO数据集中的图像。此数据集适用于微控制器视觉模型的开发和评估，特别是在需要低内存占用和高效能的边缘计算场景中，如智能家居和工业物联网应用。
The Visual Wake Words dataset was developed by the Google Research team to serve as a benchmark dataset for microcontroller-based vision applications, with a specific focus on detecting the presence of a person in an image. This dataset comprises 115,000 training and validation images derived from the publicly available COCO dataset, which has been relabeled to distinguish between the "person" and "non-person" categories. The dataset's creation involved using custom scripts to filter and annotate images from the COCO dataset. This dataset is suitable for the development and evaluation of vision models tailored for microcontrollers, particularly in edge computing scenarios requiring low memory footprint and high energy efficiency, such as smart home and industrial IoT applications.
提供机构：
谷歌研究
创建时间：
2019-06-13
