jdpoling/fjordfish_specialist_models

YOLOv9c object detectors trained on the specialist annotation splits of the FjordFish underwater imagery dataset:

Files:

• experiment_2_9c_specialist_models
  • gadids_9c # gadid specialist model
    • weights
      • best.pt # model weights here
    • model arguments and outputs
  • labrids_9c # labrids specialist model
    • weights
      • best.pt # model weights here
    • model arguments and outputs
  • sharks_9c # shark specialist model
    • weights
      • best.pt # model weights here
    • model arguments and outputs

For more information see our associated publication: link tba upon publication

If you use this model, please cite the following publication:

Poling, J. D., Perry, D., Halvorsen, K. T., Malde, K., Thormar, J., Larsen, T., & Sørdalen, T. K. (2026). Annotation strategy trade-offs for deep learning-based underwater fish detection. Ecological Informatics, 103891.