segformer-finetuned-lane-1k-steps

This model is a fine-tuned version of nvidia/segformer-b0-finetuned-cityscapes-512-1024 on the Efferbach/lane_master dataset. It achieves the following results on the evaluation set:

• Loss: 0.0548
• Mean Iou: 0.0708
• Mean Accuracy: 0.1236
• Overall Accuracy: 0.1217
• Accuracy Background: nan
• Accuracy Left: 0.1893
• Accuracy Right: 0.0578
• Iou Background: 0.0
• Iou Left: 0.1581
• Iou Right: 0.0544

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: 8
• eval_batch_size: 8
• seed: 1337
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: polynomial
• training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Left	Accuracy Right	Iou Background	Iou Left	Iou Right
0.1	1.0	308	0.0862	0.0008	0.0013	0.0012	nan	0.0025	0.0	0.0	0.0025	0.0
0.0596	2.0	616	0.0597	0.0712	0.1126	0.1132	nan	0.0940	0.1313	0.0	0.0907	0.1228
0.0506	3.0	924	0.0551	0.0682	0.1171	0.1152	nan	0.1805	0.0536	0.0	0.1539	0.0508
0.0494	3.25	1000	0.0548	0.0708	0.1236	0.1217	nan	0.1893	0.0578	0.0	0.1581	0.0544

Framework versions

• Transformers 4.28.0.dev0
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.3