The Berkeley Lab-led SYNAPS-I project is utilizing advanced AI vision models, SAM 3 and DINOv3, to automate image segmentation, significantly enhancing the efficiency of scientific research. This technology compresses the time required for 3D volume labeling from a month of manual effort to approximately 15 minutes. This effort is in line with supporting the U.S. Department of Energy’s Genesis Mission by applying innovative technology to streamline data processing in scientific discovery.
SAM 3: SAM 3 is an advanced iteration of the Segment Anything Model designed for precise pixel-level object boundary detection in images and volumes. It supports flexible, high-accuracy segmentation across diverse domains. Researchers in the SYNAPS-I project pair it with DINOv3 to achieve rapid, automated labeling of scientific 3D data.
DINOv3: DINOv3 is a self-supervised vision foundation model that excels at capturing both broad semantic understanding and detailed spatial features from images. It serves as a core component in modern computer vision applications. The SYNAPS-I project leverages its capabilities alongside SAM 3 to enable efficient automation of complex segmentation tasks.
Berkeley Lab: Lawrence Berkeley National Laboratory is a U.S. Department of Energy national laboratory that conducts multidisciplinary scientific research in physics, biology, energy, and computing. It leads advanced projects that apply cutting-edge technologies to real-world challenges. In this news, Berkeley Lab directs the SYNAPS-I project that integrates AI models to automate scientific image analysis.
SYNAPS-I project: SYNAPS-I is a Berkeley Lab-led initiative that develops and applies AI techniques to enhance scientific workflows and data interpretation. The project focuses on automating labor-intensive tasks in research environments. In the reported development, it combines DINOv3 and SAM 3 to transform image segmentation processes for 3D scientific volumes.
US Department of Energy: The U.S. Department of Energy oversees national energy policy, nuclear security, and a broad portfolio of scientific research programs across its network of national laboratories. It funds and guides missions focused on innovation and discovery. Here, the department’s Genesis Mission receives support from the SYNAPS-I project’s AI-driven tools for faster data processing.
AI in Science: AI vision models are being integrated into scientific research to automate time-consuming data labeling and segmentation workflows.
Mission Support: The SYNAPS-I effort directly advances the U.S. Department of Energy’s Genesis Mission through targeted technology applications.