segformer-b0-finetuned-segments-satellite-terrain

This model is a fine-tuned version of nvidia/mit-b0 on the ZeeeWP/terrain_map_obstacle dataset. It achieves the following results on the evaluation set:

• Loss: 0.4376
• Mean Iou: 0.3519
• Mean Accuracy: 0.7038
• Overall Accuracy: 0.7038
• Accuracy Unlabeled: nan
• Accuracy Obstacle: 0.7038
• Iou Unlabeled: 0.0
• Iou Obstacle: 0.7038

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 6e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Unlabeled	Accuracy Obstacle	Iou Unlabeled	Iou Obstacle
0.4655	3.3333	20	0.5979	0.2393	0.4786	0.4786	nan	0.4786	0.0	0.4786
0.409	6.6667	40	0.5183	0.3392	0.6784	0.6784	nan	0.6784	0.0	0.6784
0.3966	10.0	60	0.4376	0.3519	0.7038	0.7038	nan	0.7038	0.0	0.7038

Framework versions

• Transformers 4.44.1
• Pytorch 2.4.1
• Datasets 2.14.6
• Tokenizers 0.19.1