mayrajeo/marine-vessel-detection-yolov8

Model Card for YOLOv8 models for detecting marine vessels from RGB Sentinel-2 images

Model Details

Model Description

These model weights are meant for detecting marine vessels from RGB Sentinel-2 imagery.

The models are trained and validated on human-annotated marine vessel data. These annotations, information of which Sentinel-2 product they correspond to, and instructions on how to convert them into COCO or YOLO format datasets are available as GeoPackage files on Zenodo portal: https://zenodo.org/records/10046342.

• Developed by: Janne Mäyrä
• Model type: Object Detection
• Finetuned from model: YOLOv8 pretrained models

Model Sources

• Paper: http://dx.doi.org/10.2139/ssrn.4827287
• Demo: https://huggingface.co/spaces/mayrajeo/marine-vessel-detection

Uses

Direct Use

Models are trained to process 320x320 pixel patches of Sentinel-2 RGB images with 10m resolution and detect marine vessels which are upsampled into 640x640 pixels during training and inference. The models will detect targets from outside of the water areas, but those detections can be eliminated by using external datasets such as land masks.

Out-of-Scope Use

These models are not suitable for other purposes than for detecting potential marine vessels from satellite imagery.

Recommendations

The models are most accurate on more closed water areas such as denser archipelagos and close to the coast instead of open sea. Especially during windy conditions the models incorrectly detect a large number of wakes and glint as marine vessels.

How to Get Started with the Model

[https://github.com/mayrajeo/ship-detection] provides examples on how to use the models.

Training Details

Training Data

The model is trained using the following Sentinel-2 mosaics and manually annotated marine vessel data. Archipelago sea 2 and Kvarken were used as test data. Other three locations were sliced into 320x320 pixel patches. These patches were then spatially split into five equal sized folds so that each fold contained data from all timesteps and locations, and all patch locations that contained an annotated vessel in any timestep were included in the folds. In total, this dataset contained 3264 320x320 pixel image patches, of which 1974 contained annotated vessels and 1290 were background patches.

The annotations are available as GeoPackages on Zenodo portal: https://zenodo.org/doi/10.5281/zenodo.10046341

Training and validation data:

Location	Date	Number of annotations	Annotated patches	Background patches
Archipelago sea	2022-06-19	519	271	269
Archipelago sea	2022-07-21	1518	387	153
Archipelago sea	2022-08-13	1368	402	138
Gulf of Finland	2022-06-06	278	138	241
Gulf of Finland	2022-06-26	1206	269	110
Gulf of Finland	2022-07-21	971	260	119
Bothnian Bay	2022-06-27	122	81	88
Bothnian Bay	2022-07-12	162	98	71
Bothnian Bay	2022-08-28	98	68	101

Training Hyperparameters

Training configs can be found from each model directory, in the file args.yaml.

Evaluation

Testing Data, Factors & Metrics

Testing Data

Test data consists of six Sentinel-2 mosaics:

Location	Date	Number of annotations
Bothnian sea	2021-07-14	450
Bothnian sea	2022-06-24	424
Bothnian sea	2022-08-13	399
Kvarken	2022-06-17	83
Kvarken	2022-07-12	183
Kvarken	2022-08-26	88

Factors

Before evaluating, the predictions for the test set are cleaned using the following steps:

1. All prediction whose centroid points are not located on water are discarded. The water mask used contains layers jarvi (Lakes), meri (Sea) and virtavesialue (Rivers as polygon geometry) from the Topographical database by the National Land Survey of Finland. Unfortunately this also discards all points not within the Finnish borders.
2. All predictions whose centroid points are located on water rock areas are discarded. The mask is the layer vesikivikko (Water rock areas) from the Topographical database.
3. All predictions that contain an above water rock within the bounding box are discarded. The mask contains classes 38511, 38512, 38513 from the layer vesikivi in the Topographical database.
4. All predictions that contain a lighthouse or a sector light within the bounding box are discarded. Lighthouses and sector lights come from Väylävirasto data, ty_njr class ids are 1, 2, 3, 4, 5, 8
5. All predictions that are wind turbines, found in Topographical database layer tuulivoimalat
6. All predictions that are obviously too large are discarded. The prediction is defined to be "too large" if either of its edges is longer than 750 meters.

Metrics

Precision and Recall with IoU-threshold of 0.5, mAP50 and mAP.

Results

5-fold cross-validation results:

Model	('Precision', 'max')	('Precision', 'mean')	('Precision', 'min')	('Recall', 'max')	('Recall', 'mean')	('Recall', 'min')	('mAP', 'max')	('mAP', 'mean')	('mAP', 'min')	('mAP50', 'max')	('mAP50', 'mean')	('mAP50', 'min')
yolov8n	0.84688	0.82316	0.81273	0.79587	0.783928	0.76895	0.38816	0.380828	0.37637	0.84323	0.827192	0.81776
yolov8s	0.86363	0.834556	0.80835	0.83639	0.825554	0.81469	0.41098	0.406258	0.40374	0.86933	0.858632	0.84776
yolov8m	0.83592	0.827938	0.81134	0.84939	0.83238	0.80153	0.41739	0.410742	0.40496	0.87071	0.860378	0.84545
yolov8l	0.8545	0.846998	0.83973	0.84975	0.83072	0.82398	0.41694	0.411712	0.40505	0.87459	0.863334	0.85743
yolov8x	0.84028	0.834728	0.83019	0.84891	0.828462	0.80858	0.42065	0.411532	0.40231	0.86203	0.858136	0.85131

Models with best performance for test set for each model architecture:

Model	Fold	Precision	Recall	mAP50	mAP
yolov8n	1	0.820806	0.838452	0.842	0.403
yolov8s	4	0.844485	0.860565	0.865	0.422
yolov8m	4	0.858866	0.874693	0.880	0.453
yolov8l	1	0.840909	0.863636	0.862	0.443
yolov8x	1	0.855746	0.859951	0.876	0.450

Compute Infrastructure

Hardware

NVIDIA V100 GPU with 32GB of memory, hosted on computing nodes of Puhti supercomputer by CSC - IT Center for Science, Finland.

Software

Models were trained as Slurm batch jobs in Puhti.

Citation [optional]

BibTeX:

@misc{mayraMappingRecreationalMarine2024,
  type = {{{SSRN Scholarly Paper}}},
  title = {Mapping Recreational Marine Traffic from Sentinel-2 Imagery with Yolov8},
  author = {M{\"a}yr{\"a}, Janne and Virtanen, Elina and Jokinen, Ari-Pekka and Koskikala, Joni and V{\"a}kev{\"a}, Sakari and Attila, Jenni},
  year = {2024},
  doi = {10.2139/ssrn.4827287},
}

APA:

Mäyrä, Janne and Virtanen, Elina and Jokinen, Ari-Pekka and Koskikala, Joni and Väkevä, Sakari and Attila, Jenni, Mapping Recreational Marine Traffic from Sentinel-2 Imagery with Yolov8. Available at SSRN: https://ssrn.com/abstract=4827287 or http://dx.doi.org/10.2139/ssrn.4827287

Model Card Contact

Janne Mäyrä