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

This model is a fine-tuned version of nvidia/mit-b4 on the segments/sidewalk-semantic dataset. It achieves the following results on the evaluation set:

• Loss: 0.0243
• Mean Iou: 0.9582
• Mean Accuracy: 0.9792
• Overall Accuracy: 0.9965
• Accuracy Unlabeled: 0.9981
• Accuracy Numero: 0.9603
• Iou Unlabeled: 0.9963
• Iou Numero: 0.9200

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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Unlabeled	Accuracy Numero	Iou Unlabeled	Iou Numero
0.1406	5.0	20	0.1672	0.7389	0.7497	0.9790	1.0000	0.4994	0.9785	0.4993
0.045	10.0	40	0.0498	0.9398	0.9476	0.9951	0.9994	0.8958	0.9949	0.8846
0.0361	15.0	60	0.0296	0.9575	0.9811	0.9964	0.9978	0.9643	0.9963	0.9187
0.026	20.0	80	0.0243	0.9582	0.9792	0.9965	0.9981	0.9603	0.9963	0.9200

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0