谷歌公司近日发布新闻稿，介绍了小型人工智能模型 Mirasol，可以回答有关视频的问题并创造新的记录。借助 Mirasol AI 模型，该团队试图解决两个关键挑战：1.需要以高频采样同步视频和音频，但要异步处理标题和视频描述。2.视频和音频会生成大量数据，这会让模型的容量紧张。
Mirasol AI 模型是由谷歌团队开发的一款小型人工智能模型，拥有 30 亿参数，可以将多模态理解扩展到长视频。该模型的目的是为用户提供更好的视频体验，以及更准确地回答与视频相关的问题。 Mirasol 模型的诞生是为了解决视频处理中的两个难题，同步视频和音频以及异步处理标题和视频描述。同时，该模型可以处理大量数据，容量不受影响。
这款人工智能模型的发布引起了业界的高度关注。它不仅可以为用户提供更好的视频体验，还可以为视频制作和编辑提供更多的可能性。 Mirasol AI 模型将成为未来视频处理领域的重要工具，为用户带来更丰富、更准确的体验。
英文翻译：
Google has recently released a news稿，introducing a small artificial intelligence model called Mirasol，which can answer questions about videos and create new records. With Mirasol AI model，该团队试图解决两个关键挑战：1.需要以高频采样同步视频和 audio，但要异步处理标题和 video description.2.视频和 audio will generate a large amount of data，which will make the model’s capacity tight.
Mirasol AI model is developed by the Google team and has 30 billion parameters, which can expand multi-modal understanding to long videos. The purpose of this model is to provide a better video experience for users and answer questions about videos more accurately. Mirasol AI model is designed to solve the two key challenges in video processing: synchronizing video and audio and processing titles and video descriptions asynchronously. At the same time, this model can handle a large amount of data and does not have tight capacity.
This artificial intelligence model has attracted a lot of attention in the industry. It not only provides a better video experience for users but also provides more possibilities for video production and editing. Mirasol AI model will be an important tool in the future video processing field and bring more rich and accurate experiences to users.
【来源】https://www.ithome.com/0/733/030.htm
Views: 1