MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202641110592 A) filed by Vallurupalli Nageswara Rao Vignana Jyothi Institute Of Engineering And Technology on September 15, 2026, for An Artificial Intelligence-Based Automated Salt Detection System Using Deep Learning For Seismic Image Analysis And Safe Oil And Gas Exploration.
Inventor includes Dr. G. Nagaraju.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: ABSTRACT [0023] A deep-learning system is disclosed for automatic detection and pixel-wise segmentation of salt structures in seismic reflection images. Grayscale seismic slices and corresponding binary masks drawn from a public salt-identification collection are first normalized and denoised. A U-Net encoder-decoder network is then trained to produce salt probability maps; class imbalance and data scarcity are mitigated by generating additional realistic image-mask pairs with a conditional generative adversarial network. Regularization techniques including dropout and batch normalization stabilize training. After inference, probability maps are thresholded to yield binary salt masks. Quantitative evaluation demonstrates progressive gains in accuracy and Intersection-over-Union when synthetic samples are introduced, confirming that the combined architecture delivers reliable salt delineation suitable for industrial seismic interpretation workflows.
Disclaimer: Curated by HT Syndication.
