README.md

metadata

license: other
tags:
  - vision
  - image-segmentation
  - generated_from_trainer
model-index:
  - name: segformer-v-mesh-0
    results: []

segformer-v-mesh-0

This model is a fine-tuned version of nvidia/mit-b2 on the Onegafer/vehicle_segmentation dataset. It achieves the following results on the evaluation set:

• Loss: 0.0360
• Mean Iou: 0.4403
• Mean Accuracy: 0.8806
• Overall Accuracy: 0.8806
• Accuracy Background: nan
• Accuracy Windows: 0.8806
• Iou Background: 0.0
• Iou Windows: 0.8806

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: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Windows	Iou Background	Iou Windows
0.2932	0.16	20	0.3269	0.2578	0.5156	0.5156	nan	0.5156	0.0	0.5156
0.1417	0.31	40	0.1235	0.3790	0.7580	0.7580	nan	0.7580	0.0	0.7580
0.0952	0.47	60	0.1245	0.4606	0.9211	0.9211	nan	0.9211	0.0	0.9211
0.0778	0.62	80	0.0628	0.4042	0.8084	0.8084	nan	0.8084	0.0	0.8084
0.0448	0.78	100	0.0512	0.4161	0.8322	0.8322	nan	0.8322	0.0	0.8322
0.0323	0.94	120	0.0435	0.4167	0.8334	0.8334	nan	0.8334	0.0	0.8334
0.0337	1.09	140	0.0405	0.4131	0.8262	0.8262	nan	0.8262	0.0	0.8262
0.0586	1.25	160	0.0409	0.4509	0.9017	0.9017	nan	0.9017	0.0	0.9017
0.0591	1.41	180	0.0404	0.4310	0.8620	0.8620	nan	0.8620	0.0	0.8620
0.0953	1.56	200	0.0386	0.4366	0.8732	0.8732	nan	0.8732	0.0	0.8732
0.0607	1.72	220	0.0374	0.4414	0.8828	0.8828	nan	0.8828	0.0	0.8828
0.0387	1.88	240	0.0360	0.4403	0.8806	0.8806	nan	0.8806	0.0	0.8806

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.3