Sam 2: The next generation of Meta Segment Anything Model
Meta has launched SAM 2, a real-time object segmentation model for images and videos, enhancing accuracy and reducing interaction time. It supports diverse applications and is available under an Apache 2.0 license.
Read original article
Meta has introduced SAM 2, the next generation of its Segment Anything Model, which now supports real-time object segmentation in both images and videos. This unified model is designed to achieve state-of-the-art performance and can segment any object, even those it has not previously encountered, without requiring custom adaptation. SAM 2 is being released under an Apache 2.0 license, allowing developers to utilize the model freely. Alongside the model, Meta is sharing the SA-V dataset, which contains approximately 51,000 real-world videos and over 600,000 spatio-temporal masks, significantly expanding the resources available for video segmentation tasks.
The model enhances the capabilities of its predecessor, SAM, by improving segmentation accuracy and reducing interaction time by threefold. SAM 2's architecture incorporates a memory mechanism that allows it to maintain context across video frames, addressing challenges such as object motion and occlusion. This advancement enables applications in various fields, including video editing, scientific research, and autonomous vehicles.
Meta emphasizes its commitment to open science by providing the SAM 2 code, weights, and evaluation tools, encouraging the AI community to explore new use cases. The model's potential applications range from creative video effects to aiding in medical procedures and environmental research. With SAM 2, Meta aims to further revolutionize the field of computer vision and inspire innovative solutions across multiple industries.
Related
Video annotator: a framework for efficiently building video classifiers
The Netflix Technology Blog presents the Video Annotator (VA) framework for efficient video classifier creation. VA integrates vision-language models, active learning, and user validation, outperforming baseline methods with an 8.3 point Average Precision improvement.
Meta won't release its multimodal Llama AI model in the EU
Meta will not release its Llama AI model in the EU due to regulatory concerns, impacting European companies. Apple also considers excluding the EU from its AI rollout. This decision poses challenges for companies worldwide.
Big tech wants to make AI cost nothing
Meta has open-sourced its Llama 3.1 language model for organizations with fewer than 700 million users, aiming to enhance its public image and increase product demand amid rising AI infrastructure costs.
TreeSeg: Hierarchical Topic Segmentation of Large Transcripts
Augmend is creating a platform to automate tribal knowledge for development teams, featuring the TreeSeg algorithm, which segments session data into chapters by analyzing audio transcriptions and semantic actions.
Sam 2: Segment Anything in Images and Videos
The GitHub repository for Segment Anything Model 2 (SAM 2) by Facebook Research enhances visual segmentation with real-time video processing, a large dataset, and APIs for image and video predictions.
1 commentsRelated
Video annotator: a framework for efficiently building video classifiers
The Netflix Technology Blog presents the Video Annotator (VA) framework for efficient video classifier creation. VA integrates vision-language models, active learning, and user validation, outperforming baseline methods with an 8.3 point Average Precision improvement.
Meta won't release its multimodal Llama AI model in the EU
Meta will not release its Llama AI model in the EU due to regulatory concerns, impacting European companies. Apple also considers excluding the EU from its AI rollout. This decision poses challenges for companies worldwide.
Big tech wants to make AI cost nothing
Meta has open-sourced its Llama 3.1 language model for organizations with fewer than 700 million users, aiming to enhance its public image and increase product demand amid rising AI infrastructure costs.
TreeSeg: Hierarchical Topic Segmentation of Large Transcripts
Augmend is creating a platform to automate tribal knowledge for development teams, featuring the TreeSeg algorithm, which segments session data into chapters by analyzing audio transcriptions and semantic actions.
Sam 2: Segment Anything in Images and Videos
The GitHub repository for Segment Anything Model 2 (SAM 2) by Facebook Research enhances visual segmentation with real-time video processing, a large dataset, and APIs for image and video predictions.