Meta, the company founded by Mark Zuckerberg, has launched its own AI model called the “Segment Anything Model” (SAM), capable of detecting and identifying objects in images and videos, along with generating complex images and videos from text.
As reported by Reuters, the new technology is described in a Meta blog post, where a demonstration of SAM shows its ability to recognize cats in a photo when prompted by typing the word “cat.”
The emergence of Meta’s AI technology follows the rising trend of AI platforms, including Microsoft-backed ChatGPT, which recently faced backlash in Italy over a privacy breach that led to a temporary suspension of the service. Despite the controversy, the popular cartoon “South Park” even used ChatGPT to write part of a recent episode, showing the growing influence of AI in various industries.
Meta unveils new AI model as CEO Mark Zuckerberg pushes to incorporate generative artificial intelligence into Meta's apps in 2023 https://t.co/P9lEWCberd pic.twitter.com/YSHepMrEVn
— Reuters (@Reuters) April 7, 2023
Meta’s new AI capabilities are set to revolutionize the way we generate images and videos. The tech giant has announced “Make-A-Scene,” an application that allows users to create images with text prompts, going beyond just text-to-image generation to “Scene-Based Text-to-Image Generation with Human Priors.”
Make-A-Scene can enhance and upgrade existing images and create new ones from scratch, as demonstrated in a published video showing a story created using simple sketches and text prompts.
Meta’s new AI program, SAM, is capable of identifying objects in images and videos, creating complex images and videos from text prompts. According to a Meta blog post, SAM can draw boxes around specific objects in images such as cats.
Meta has even compared its programs with competitors, demonstrating its ability to create cartoon-like sketches, realistic images, and even imaginary concepts such as “a car with triangular wheels.” With Meta’s new AI capabilities, the possibilities for generating high-quality, custom-made images and videos are limitless.
The “Make-A-Video” studio empowers users to create dynamic videos by combining text and one or more images, or by modifying existing videos.
The first video sample is a thrilling clip of a dog sporting a superhero costume and a red cape as it soars through the sky.
The platform offers various surreal video types, such as a teddy bear painting a portrait, a young couple strolling in heavy rain under the realistic category, and a stylized video of a table positioned by a window, with sunlight streaming through, illuminating a pile of books.
The SAM model and dataset will be accessible for download under a non-commercial license that enables the AI platform to be used for research purposes exclusively.
