SAM 2 and SA-V at a glance: Meta simplifies visual editing
On your own blog explains Metawhat the new generation of the Segment Anything Model has to offer. Above all, the increased precision in object marking in images and videos stands out.
Meta particularly wanted to optimise the application possibilities in the moving image sector:
SAM was able to learn a general notion of what objects are in images. However, images are only a static snapshot of the dynamic real world in which visual segments can exhibit complex motion. Many important real-world use cases require accurate object segmentation in video data, for example in mixed reality, robotics, autonomous vehicles, and video editing. We believe that a universal segmentation model should be applicable to both images and video (…).
In parallel with SAM 2, Meta launched the SA-V dataset with over 51,000 real-world videos. This is available under the CC BY 4.0 license and offers around 53 times more annotations than the largest video dataset for object marking to date.
Like the original SAM, SAM 2 can be applied out of the box to a diverse range of real-world use cases and we’re excited to see what developers build with it.
Try the demo https://t.co/lDmWrzA2tZ
Get the code https://t.co/qWCPI43A6I— AI at Meta (@AIatMeta) July 29, 2024
On the Meta Blog you can find out more about the technical details, application options, but also limitations of SAM 2. In fast-paced scenarios, object marking can sometimes get lost.
One potential application scenario, however, is the integration into AI Glasses – which Meta itself also offers – whereby individual objects in the user’s view could be immediately recognized and classified.
After increase in sales and user growth:
Meta wants to push AI in all services
Source: onlinemarketing.de
