segformer-b0-finetuned-segments-sidewalk-2

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

• Loss: 0.3429
• Mean Iou: 0.8143
• Mean Accuracy: 0.9007
• Overall Accuracy: 0.9061
• Per Category Iou: [0.8822819675417668, 0.7774253195321242, 0.7832033563111727]
• Per Category Accuracy: [0.9319684170082266, 0.8657193844491432, 0.9044945609610779]

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: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Per Category Iou	Per Category Accuracy
0.7949	0.5	20	0.8960	0.7129	0.8533	0.8427	[0.7978191889735743, 0.6994730230171242, 0.6413103816527537]	[0.826874349660607, 0.8237981626592454, 0.9091007880329902]
0.4881	1.0	40	0.6195	0.7364	0.8610	0.8552	[0.8041892620489134, 0.6981663805103046, 0.7069887055480671]	[0.8308827565320059, 0.887905283397269, 0.8642919506720577]
0.3115	1.5	60	0.4767	0.7352	0.8536	0.8588	[0.8276338695141907, 0.7016825436162023, 0.6763414045904438]	[0.8633649830215921, 0.8776778472775076, 0.8196451790592317]
0.5863	2.0	80	0.4895	0.7543	0.8748	0.8668	[0.8156517914197925, 0.7259786638902507, 0.7213518497027839]	[0.8402281798360435, 0.8932153836673491, 0.8909222571543128]
0.5182	2.5	100	0.4058	0.7904	0.8866	0.8919	[0.860991170688589, 0.7583876635226005, 0.7518265397248736]	[0.9088903949664655, 0.8761789935147187, 0.8746304338865427]
0.4755	3.0	120	0.3683	0.7896	0.8861	0.8895	[0.8547537413009911, 0.7465075384127533, 0.7674680941571024]	[0.8979683913158062, 0.8865259395690547, 0.8738060532025316]
0.6616	3.5	140	0.3697	0.7915	0.8874	0.8898	[0.8551700094228354, 0.7431970428539307, 0.7761922571371438]	[0.8899387313627766, 0.903193218309171, 0.8690639906770039]
0.5087	4.0	160	0.3367	0.8061	0.8987	0.8987	[0.8640367246398447, 0.7643869962764198, 0.7899951558528526]	[0.9012200396208266, 0.8918889478830869, 0.902900133774502]
0.5478	4.5	180	0.3297	0.8131	0.8991	0.9040	[0.8775309087721331, 0.7692790103652185, 0.792538025793261]	[0.9196387801394476, 0.8895118205906903, 0.8882327151727265]
0.389	5.0	200	0.3429	0.8143	0.9007	0.9061	[0.8822819675417668, 0.7774253195321242, 0.7832033563111727]	[0.9319684170082266, 0.8657193844491432, 0.9044945609610779]

Framework versions

• Transformers 4.20.1
• Pytorch 1.12.0+cu113
• Datasets 2.3.2
• Tokenizers 0.12.1