Maritime Vessel Recognition (SSL Pretrained Backbones)
The checkpoints below are teacher backbones from self-supervised pretraining on the in-distribution maritime vessel dataset. Each .pth file is a teacher_checkpoint.pth state dict.
Models
| Checkpoint Name | Architecture | SSL Method | Backbone Size |
|---|---|---|---|
| vit-b-16-dinov2.pth | ViT-B/16 | DINOv2 | 416 MB |
| vit-l-16-dinov2.pth | ViT-L/16 | DINOv2 | 1.3 GB |
| vit-h-14-dinov2.pth | ViT-H/14 | DINOv2 | 2.5 GB |
| vit-l-16-franca.pth | ViT-L/16 | Franca | 2.0 GB |
| vit-l-16-simdinov2.pth | ViT-L/16 | SimDINOv2 | 1.2 GB |