|
--- |
|
license: cc-by-4.0 |
|
tags: |
|
- ocean |
|
- benthic |
|
- object-detection |
|
--- |
|
|
|
# FathomNet2023 Baseline Model |
|
|
|
## Model Details |
|
- Trained by researchers at [Monterey Bay Aquarium Research Institute](https://www.mbari.org/) (MBARI) as a baseline for the [FathomNet2023 Competition](https://www.kaggle.com/competitions/fathomnet-out-of-sample-detection/overview) presented with the [Fine Grained Visual Categorization workshop](https://sites.google.com/view/fgvc10/home) at CVPR 2023. |
|
- [Ultralytics YOLOv8.0.117](https://github.com/ultralytics/ultralytics/pull/3145) |
|
- Object detection |
|
- Fine tuned yolov8m to detect 290 fine grained taxonmic categories of benthic animals in the Greater Monterey Bay Area off the coast of Central California. |
|
|
|
## Intended Use |
|
- Make detections on images collect on the sea floor in the Monterey Bay Area. |
|
|
|
## Factors |
|
- Distribution shifts related to sampling platform, camera parameters, illumination, and deployment environment are expected to impact model performance. |
|
- Evaluation was performed on an IID subset of available training data. |
|
- Data to test out of distribution performance can be found on the [competition Kaggle page](https://www.kaggle.com/competitions/fathomnet-out-of-sample-detection/overview). |
|
|
|
## Metrics |
|
- [Precision-Recall curve](https://huggingface.co/FathomNet/MBARI-midwater-supercategory-detector/blob/main/plots/PR_curve.png) and [per class accuracy]((https://huggingface.co/FathomNet/MBARI-midwater-supercategory-detector/blob/main/plots/confusion_matrix.png)) were evaluated at test time. |
|
- [email protected] = 0.33515 |
|
- Performance is quite variable depending on the target organism even when testing on in-distribution data. |
|
|
|
## Training and Evaluation Data |
|
- Training data is the [FathomNet2023 competition split](https://www.kaggle.com/competitions/fathomnet-out-of-sample-detection/overview) and internal MBARI data |
|
- Class labels have a [long tail and localizations occur throughout the frame](). |
|
|
|
## Deployment |
|
In an environment running YOLOv8: |
|
|
|
``` |
|
python classify/predict.py --weights best.pt --data data/images/ |
|
``` |
|
|