segformer-b0-finetuned-segments-sidewalk-2

This model is a fine-tuned version of nvidia/mit-b0 on the erikhenriksson1/test_skin_segmentation dataset. It achieves the following results on the evaluation set:

• Loss: 0.1985
• Mean Iou: 0.6571
• Mean Accuracy: 0.6626
• Overall Accuracy: 0.9937
• Accuracy Unlabeled: nan
• Accuracy Skin: 0.9939
• Accuracy Mark of interest: 0.0
• Accuracy Non-skin: 0.9939
• Iou Unlabeled: nan
• Iou Skin: 0.9801
• Iou Mark of interest: 0.0
• Iou Non-skin: 0.9911

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 Skin	Accuracy Mark of interest	Accuracy Non-skin	Iou Unlabeled	Iou Skin	Iou Mark of interest	Iou Non-skin
0.6293	5.0	20	0.9612	0.5999	0.6435	0.9522	nan	1.0	0.0	0.9306	nan	0.8691	0.0	0.9306
0.4646	10.0	40	0.4611	0.6321	0.6552	0.9765	nan	0.9990	0.0	0.9665	nan	0.9304	0.0	0.9660
0.3595	15.0	60	0.3382	0.6423	0.6578	0.9837	nan	0.9944	0.0	0.9791	nan	0.9505	0.0	0.9766
0.3065	20.0	80	0.3041	0.6461	0.6592	0.9863	nan	0.9951	0.0	0.9826	nan	0.9581	0.0	0.9804
0.2874	25.0	100	0.2659	0.6506	0.6604	0.9893	nan	0.9934	0.0	0.9877	nan	0.9669	0.0	0.9848
0.2319	30.0	120	0.2461	0.6522	0.6614	0.9904	nan	0.9961	0.0	0.9881	nan	0.9703	0.0	0.9864
0.2396	35.0	140	0.2279	0.6565	0.6626	0.9933	nan	0.9952	0.0	0.9927	nan	0.9790	0.0	0.9906
0.2215	40.0	160	0.2099	0.6564	0.6624	0.9932	nan	0.9939	0.0	0.9932	nan	0.9788	0.0	0.9904
0.2128	45.0	180	0.2021	0.6565	0.6626	0.9933	nan	0.9948	0.0	0.9929	nan	0.9789	0.0	0.9905
0.1855	50.0	200	0.1985	0.6571	0.6626	0.9937	nan	0.9939	0.0	0.9939	nan	0.9801	0.0	0.9911

Framework versions

• Transformers 4.39.1
• Pytorch 2.2.1
• Datasets 2.18.0
• Tokenizers 0.15.2