Faults Handler

Faults—fractures within the Earth's crust—play a crucial role in the migration and trapping of hydrocarbons. They can either facilitate the movement of hydrocarbons into reservoirs or act as barriers that impede their flow. Accurate identification and interpretation of these faults are essential for geologists involved in oil and gas exploration. In three-dimensional (3D) seismic data, faults are relatively easier to interpret due to the continuous spatial coverage across the area of interest. However, in frontier exploration areas, geologists often have access only to two-dimensional (2D) seismic data, which consists of discrete inline and crossline sections. This lack of continuity makes it challenging to identify and correlate the same fault structures across different 2D lines, especially in regions with complex faulting.

This project aims to address these challenges by developing open-source web application for effective fault detection in 2D seismic images using AI model. Currently, the application run using YOLOv11 segmentation model

Setup

1. Clone this repository

   git clone https://github.com/haizadtarik/faults-handler.git
   cd faults-handler
   

2. Install necessary dependencies

   pip install -r requirements.txt
   cd client
   npm install --force
   

Run application

1. Launch server

   python src/server.py
   

2. On new terminal, launch client application

   cd client
   npm run dev
   

3. Go to http://localhost:3000 and upload your 2D seismic images to identify the faults

Train on Your Own Data

Data need to be in yolo format. For more details, refer here

To generate labels file from mask images, use function create_yolo_segmentation_labels_from_directory in util.py

To train modify configuration in config/seismic_data.yaml and train.py and run:

python src/train.py