谷歌近日宣布开源其先进的文件识别技术Magika，该技术能以毫秒级的速度准确识别文件格式和内容类型。 Magika 背后的深度学习模型经过高度优化，即使在 CPU 上运行也能实现如此高效的识别速度。在百万级别的文件测试中，Magika的准确率超过99%。
这项技术不仅提高了工作效率，也极大地方便了用户对大量文件的管理。Magika 的开源意味着更多的开发者和企业可以利用这一技术进行二次开发或集成到现有系统中，以提升文件处理的速度和准确性。
据谷歌介绍，Magika 的源代码已托管至GitHub，供全球开发者共同使用和改进。开源社区的加入有望进一步提升 Magika 的性能和适用范围。
英文标题：Google Open Sources Magika: Millisecond-Level File Type Recognition
英文关键词：Google, Magika, File Recognition
英文新闻内容：
Google has recently announced the open-sourcing of its advanced file recognition technology, Magika, which can accurately identify file formats and content types at millisecond speeds. The deep learning model behind Magika has been highly optimized to achieve such efficient recognition even on CPUs. In tests involving millions of files, Magika achieved an accuracy rate of over 99%.
This technology not only enhances work efficiency but also greatly facilitates users in managing a large volume of files. The open-sourcing of Magika means that more developers and enterprises can utilize this technology for secondary development or integration into existing systems, thus improving the speed and accuracy of file handling.
According to Google, the source code of Magika has been hosted on GitHub for global developers to use, improve, and contribute to. The involvement of the open-source community is expected to further enhance Magika’s performance and application scope.
【来源】https://www.ithome.com/0/750/474.htm
Views: 1