1887

Abstract

Summary

Artificial Intelligence, particularly Computer Vision, has revolutionized image segmentation, enabling the detection of various objects. In the energy sector, AI interprets geological objects in different kind of seismic data. However, deploying such solutions efficiently at scale is challenging due to different data sizes, types, and segmentation purposes. Additionally, in a Cloud environment, GPU resources may be limited, with multiple users requesting access simultaneously.

To address this, we optimized our solution focusing on maximizing throughput and resource occupancy for multiple users. By employing tiling techniques and dynamic load balancing, we ensured computational efficiency and minimized waiting times. Moreover, the Multi-Instance GPU (MIG) configuration was enabled, dividing GPU resources into independent partitions, virtually adding more GPUs.

We illustrated this strategy in a multi-user Cloud environment, proving its effectiveness.

Loading

Article metrics loading...

/content/papers/10.3997/2214-4609.2025643017
2025-10-06
2026-02-15
Loading full text...

Full text loading...

References

  1. Boillot, L., Pivot, F. and Klein, F. [2024a] Synthetic label masks mapped on ocean satellite background for oil seepage detection. International Conference on Scientific Computing and Machine Learning
    [Google Scholar]
  2. Boillot, L. and Shreif, M. [2024b] 3D Channel Detection Workflow Using Spectral Decomposition for Colorized Image Segmentation. Fourth EAGE Digitalization Conference & Exhibition
    [Google Scholar]
  3. Boillot, L., Thouvenot, A., Kuhn de Chizelle, J. and Guillon, S. [2023] Stratigraphic constraint for deep learning image segmentation of geological units. SEG International Exposition and Annual Meeting
    [Google Scholar]
  4. Goutorbe, P., Joao, C., Scherrier, H. and Umoren, N. [2021] First Operational AI Fault Interpretation Study Performed at Total. 82nd EAGE Annual Conference & Exhibition
    [Google Scholar]
  5. Krajnc-GiroudA. and BoillotL. [2023] Unlocking fast mini basins exploration workflows using deep learning salt detection. 84th EAGE Annual Conference & Exhibition
    [Google Scholar]
  6. Li, B., Gadepally, V., Samsi, S., and Tiwari, D. [2022] Characterizing multi-instance gpu for machine learning workloads. 2022 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW), pp. 724–731.
    [Google Scholar]
  7. Pichat, A., Célini, N., Schueller, S., Mengus, J.M., Gout, C., Delahaye, S. and Callot, J.P. [2025] Intra-salt deformation of K- Mg salt layers in diapiric structures, an analogue modelling approach. Tectonophysics, 898, p.230635.
    [Google Scholar]
  8. Regaieg, M., Varloteaux, C., Farhana Faisal, T. and ElAbid, Z. [2023] Towards Large-Scale DRP Simulations: Generation of Large Super-Resolution images and Extraction of Large Pore Network Models. Transport in Porous Media, 147(2), pp.375–399.
    [Google Scholar]
  9. Settles, B. [1995] Active Learning Literature Survey. Science, 10(3), pp.237–304.
    [Google Scholar]
  10. Yu, F., Wang, D., Shangguan, L., Zhang, M., Liu, C., and Chen, X. [2022] A survey of multi-tenant deep learning inference on gpu. arXiv preprint arXiv:2203.09040.
    [Google Scholar]
  11. Zhang, H., Li, Y., Xiao, W., Huang, Y., Di, X., Yin, J., SeeS., LuoY., LauC. T., and You, Y. [2023] MIGPerf: A Comprehensive Benchmark for Deep Learning Training and Inference Workloads on Multi-Instance GPUs. arXiv preprint arXiv:2301.00407.
    [Google Scholar]
/content/papers/10.3997/2214-4609.2025643017
Loading
/content/papers/10.3997/2214-4609.2025643017
Loading

Data & Media loading...

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error