EdgeSAM: Prompt-In-the-Loop Distillation for On-Device Deployment of SAM Gradio demo is out on Hugging Face demo: huggingface.co/spaces/chongzh… run with docker: huggingface.co/spaces/chongzh… duplicate space with private gpu and no queue: huggingface.co/spaces/chongzh… This paper presents

mmlab-ntu presents Open-Vocabulary SAM Segment and Recognize Twenty-thousand Classes Interactively paper page: huggingface.co/papers/2401.02… Open-Vocabulary SAM extends SAM's segmentation capabilities with CLIP-like real-world recognition, while significantly reducing computational

We turned our method, rejected by CVPR and ECCV, into the iOS app "Cutcha". EdgeSAM, our fast Segment Anything Model, runs at over 30 FPS on an iPhone 14. Enjoy intuitive one-touch object selection and precise editing—all processed locally on your device. No cloud needed!

Efficient Track Anything is accepted to ICCV 2025! See you in Hawaii!

EdgeTAM, real-time segment tracker by Meta is now in Hugging Face transformers with Apache-2.0 license 🔥 > 22x faster than SAM2, processes 16 FPS on iPhone 15 Pro Max with no quantization > supports single/multiple/refined point prompting, bounding box prompts

🚀 Synvo AI We’re excited to open-source Local-Cocoa, an on-device AI assistant that turns your files into semantic memory while keeping everything private. Built in collaboration with MMLab@NTU, and we warmly welcome contributors to help push it further! 🔧✨ 👉

🚨 VideoAuto-R1 from Meta!🚨 🚀Excited to share our VideoAuto-R1: the first video understanding framework that only reasons when necessary. Our VideoAuto-R1 produces dramatically better results than video reasoning models such as Video-R1 by +10% on both video perception and