Reproduce “Adapting a Vision Foundation Model for Fracture Characterization in Acoustic Image Logs” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Guangyu Wang, Xiaodong Ma, Xinming Wu
The authors adapt DINOv3, pretrained on satellite imagery, to classify acoustic borehole image-log pixels as background, natural fractures, or drilling-induced tensile fractures. Geometric processing groups predicted pixels into traces and fits sinusoids or lines to estimate measurements. On 20 held-out images, mean intersection over union is 0.624, versus 0.545 for U-Net; this measures segmentation overlap, not verified fracture picks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.