近日，全球瞩目的AI领域新星Omer Bar-Tal宣布加入初创公司Pika，并担任创始科学家。Omer Bar-Tal在谷歌研究院的9个月里，以Student Researcher的身份进行研究，最终在7个月内完成了Lumiere模型的开发，并以其共同一作的身份发布。彼时，他尚未完成硕士学位。
Lumiere模型的一大创新在于提出了时空U-Net（STU-Net）架构，该架构能够在空间和时间两个维度同时进行视频的下采样和上采样，从而在网络的中间层得到视频的压缩时空表示。经过对3000万视频的学习，Lumiere模型能够支持多种功能，包括文生视频、视频编辑修复、图片转视频、视频风格化等。
Omer Bar-Tal的加盟标志着Pika公司将在视频生成技术上迈向新的高度，同时也预示着AI技术在视频内容创作领域的广阔应用前景。这一消息无疑将对整个AI行业产生深远影响，并对视频生成技术的未来发展方向产生重要指引。
英文翻译内容：
Title: Google AI Star Omer Bar-Tal Joins Pika as Founding Scientist, Leading the Way in Video Generation Technology
Keywords: AI Video Generation, Pika, Lumiere Model
News content:
Recently, Omer Bar-Tal, a rising star in the AI field, announced his joining the startup company Pika as a founding scientist. During his 9-month tenure at Google Research as a Student Researcher, Bar-Tal completed the development of the Lumiere model in just 7 months and released it as a co-first author. At the time, he had not yet completed his master’s degree.
The Lumiere model’s significant innovation lies in its proposed Spatio-Temporal U-Net (STU-Net) architecture, which is capable of simultaneously downsampling and upsampling videos in both spatial and temporal dimensions, thereby obtaining a compressed spatio-temporal representation of the video at the middle layer of the network. After learning from 30 million videos, the Lumiere model supports various functions, including text-to-video generation, video editing and repair, image-to-video conversion, and video stylization.
Bar-Tal’s joining Pika signifies that the company is poised to reach new heights in video generation technology and also indicates the vast potential of AI technology in the field of video content creation. This news is undoubtedly bound to have a profound impact on the entire AI industry and provide important guidance on the future direction of video generation technology development.
【来源】https://www.qbitai.com/2024/02/123274.html
Views: 1