segformer-b0-finetuned-segments-sidewalk-oct-22

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

• Loss: 0.9792
• Mean Iou: 0.0964
• Mean Accuracy: 0.2300
• Overall Accuracy: 0.2636
• Accuracy Unlabeled: nan
• Accuracy Tank: 0.4595
• Accuracy Artillery: nan
• Accuracy Apc: 0.0005
• Iou Unlabeled: 0.0
• Iou Tank: 0.2885
• Iou Artillery: nan
• Iou Apc: 0.0005

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: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Unlabeled	Accuracy Tank	Accuracy Artillery	Accuracy Apc	Iou Unlabeled	Iou Tank	Iou Artillery	Iou Apc
0.9221	10.0	20	1.2904	0.1235	0.4284	0.4911	nan	0.8568	nan	0.0	0.0	0.4942	0.0	0.0
0.7631	20.0	40	1.0716	0.1194	0.2947	0.3378	nan	0.5887	nan	0.0008	0.0	0.3575	nan	0.0008
0.6631	30.0	60	0.9722	0.1060	0.2463	0.2818	nan	0.4888	nan	0.0038	0.0	0.3143	nan	0.0038
0.5846	40.0	80	0.9368	0.0827	0.1925	0.2207	nan	0.3850	nan	0.0	0.0	0.2481	nan	0.0
0.5662	50.0	100	0.9792	0.0964	0.2300	0.2636	nan	0.4595	nan	0.0005	0.0	0.2885	nan	0.0005

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2