Instructions to use facebook/sam3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use facebook/sam3 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("mask-generation", model="facebook/sam3")# Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("facebook/sam3") model = AutoModelForMultimodalLM.from_pretrained("facebook/sam3") - Notebooks
- Google Colab
- Kaggle
Request for SAM3
Hello Meta SAM3 Team,
My name is Shan Cao, and I am a PhD student in Artificial Intelligence at Jilin University.
I recently requested access to the facebook/sam3 repository on Hugging Face, but my request was unfortunately rejected. I would like to kindly ask whether my application could be reconsidered.
My research focuses on computer vision, image segmentation, and cultural heritage reconstruction. In particular, I am currently working on image segmentation and fragment reassembly tasks involving archaeological and historical artifacts. I am interested in evaluating SAM3's capabilities for promptable concept segmentation and comparing its performance with previous versions of the Segment Anything family.
The model will be used strictly for academic and non-commercial research purposes, including experimental evaluation, reproducibility studies, and scientific publications. I will fully comply with the SAM3 license terms, Meta policies, and all applicable usage restrictions. I do not intend to redistribute the model weights or use the model for commercial purposes.
I have also updated my Hugging Face profile information to better reflect my academic affiliation and research background.
Thank you very much for your time and consideration. I would greatly appreciate the opportunity to use SAM3 in my research.
Best regards,
Shan Cao
PhD Student
Jilin University