Seismic First Break Point Data Labeling and Refinement (DeepFBT Project)
I participated in the process of labeling seismic data for first break point picking and result refinement. The workflow involved reading raw SEG-Y seismic data, performing both manual and automatic picking of first arrivals, and integrating user corrections to improve label quality. Automated label refinement was achieved through local window extremum search and post-processing. • Labeled seismic waveform data to identify and mark first break points. • Used manual annotation and AI-assisted methods for hybrid labeling. • Employed custom tools built with C++17, Qt, and OpenCV for data visualization and interaction. • Ensured results were refined through iterative user and automated corrections.