Meta, formerly known as Facebook, has launched an AI tool that allows users to identify and separate specific items in an image. The tool, called Segment Anything, can individually identify the pixels that make up an image and separate one or more items from the rest of the image with just a few clicks. While the tool is still in demo mode, it has been trained on over 11 million images and has already shown impressive capabilities in handling various types of images that were not seen during training, such as microscopy, underwater photos, and ego-centric images. Segment Anything has been made available for the research community under a permissive open license, Apache 2.0, that can be accessed through the Segment Anything Github.
Meta coming in hot with SAM— Nick St. Pierre (@nickfloats) April 5, 2023
Segment Anything Model (SAM) is a promptable segmentation system. It can "cut out" any object, in any image, with a single click.
Masks can be tracked in videos, enable image editing apps, and even be lifted to 3D
🧵Quick tour and test pic.twitter.com/YC0JSWYy9X
The use of AI in image segmentation raises concerns about privacy violations. Data collection must be done transparently and with the individual’s full consent. The use of AI for facial recognition without express consent can potentially violate privacy laws. Companies should avoid sharing facial data with third parties unless the individual has consented, and any sharing must adhere to privacy law provisions. Meta is taking steps to ensure the privacy of its users. The company has employed various privacy-preserving techniques, such as blurring faces and other personally identifying information, and users can report offensive content to Meta by sending an email to firstname.lastname@example.org with the id of the image, and the company will remove it from the dataset.
Meta’s decision to pivot from its metaverse plans to focus on other products, including AI, came after the company laid off over 10,000 workers after ending its Instagram NFT project. Meta’s AI ambitions are significant, with the company intending to use its proprietary generative AI to create ads for companies by the end of this year. The company plans for artificial intelligence to design those ads for brands that advertise on Meta’s social platforms. The AI system will be able to generate text, images, or other media in response to prompts.
Many companies face challenges in obtaining access to large-scale training data, making social media networks the best source of training data. However, companies should include a machine learning clause that informs users how their data is being used and whether they can opt-out of future training models. The opt-in default setting of many companies may change to opt-out in the future. Concerns about AI and privacy have led several world governments to express concerns and open investigations into the technology and what it means for user privacy and safety after the launch of OpenAI’s ChatGPT. Italy has already banned the popular chatbot.
Segment Anything is still in the research phase, and there are no plans to use it in production. However, the technology has the potential to be a game-changer in the field of computer vision, with many applications, from analyzing scientific imagery to editing photos. The technology could be used to create more personalized and interactive experiences in virtual and augmented reality, paving the way for a more immersive metaverse in the future. Meta’s shift to AI may mark the beginning of a new era for the company and the tech industry as a whole.