segformer-b3-fashion-finetuned-polo-segments-v1.5

This model is a fine-tuned version of sayeed99/segformer-b3-fashion on the sshk/polo-badges-segmentation dataset. It achieves the following results on the evaluation set:

• Loss: 0.1007
• Mean Iou: 0.8404
• Mean Accuracy: 0.9136
• Overall Accuracy: 0.9704
• Accuracy Unlabeled: nan
• Accuracy Collar: 0.8876
• Accuracy Polo: 0.9746
• Accuracy Lines-cuff: 0.7358
• Accuracy Lines-chest: 0.9360
• Accuracy Human: 0.9631
• Accuracy Background: 0.9848
• Accuracy Tape: nan
• Iou Unlabeled: nan
• Iou Collar: 0.7360
• Iou Polo: 0.9428
• Iou Lines-cuff: 0.6178
• Iou Lines-chest: 0.8353
• Iou Human: 0.9386
• Iou Background: 0.9718
• Iou Tape: nan

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: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Unlabeled	Accuracy Collar	Accuracy Polo	Accuracy Lines-cuff	Accuracy Lines-chest	Accuracy Human	Accuracy Background	Accuracy Tape	Iou Unlabeled	Iou Collar	Iou Polo	Iou Lines-cuff	Iou Lines-chest	Iou Human	Iou Background	Iou Tape
0.1679	2.2222	20	0.2031	0.5532	0.5985	0.9492	nan	0.6856	0.9707	0.0	0.0022	0.9482	0.9843	nan	nan	0.5491	0.8882	0.0	0.0022	0.9192	0.9604	nan
0.0921	4.4444	40	0.1359	0.7103	0.7618	0.9631	nan	0.8586	0.9739	0.1373	0.6598	0.9577	0.9835	nan	nan	0.6786	0.9244	0.1373	0.6217	0.9305	0.9691	nan
0.0603	6.6667	60	0.1166	0.8147	0.8651	0.9672	nan	0.8436	0.9795	0.6385	0.7867	0.9586	0.9837	nan	nan	0.7114	0.9315	0.5955	0.7446	0.9352	0.9700	nan
0.0581	8.8889	80	0.1121	0.8185	0.8809	0.9677	nan	0.8363	0.9767	0.6995	0.8279	0.9594	0.9857	nan	nan	0.7091	0.9336	0.6009	0.7611	0.9357	0.9709	nan
0.0445	11.1111	100	0.1047	0.8317	0.9033	0.9699	nan	0.8719	0.9687	0.7198	0.9070	0.9686	0.9836	nan	nan	0.7263	0.9403	0.6081	0.8045	0.9390	0.9721	nan
0.0456	13.3333	120	0.1055	0.8342	0.9151	0.9694	nan	0.8931	0.9687	0.7391	0.9402	0.9614	0.9878	nan	nan	0.7285	0.9405	0.6102	0.8184	0.9371	0.9708	nan
0.0443	15.5556	140	0.1034	0.8349	0.9039	0.9700	nan	0.8740	0.9742	0.7208	0.9059	0.9636	0.9851	nan	nan	0.7324	0.9411	0.6091	0.8166	0.9384	0.9717	nan
0.0475	17.7778	160	0.1032	0.8384	0.9139	0.9699	nan	0.8885	0.9738	0.7383	0.9356	0.9604	0.9868	nan	nan	0.7341	0.9409	0.6160	0.8300	0.9377	0.9717	nan
0.0411	20.0	180	0.1018	0.8403	0.9150	0.9702	nan	0.8911	0.9770	0.7389	0.9378	0.9592	0.9862	nan	nan	0.7362	0.9417	0.6194	0.8346	0.9383	0.9716	nan
0.0345	22.2222	200	0.1003	0.8397	0.9112	0.9704	nan	0.8885	0.9768	0.7359	0.9201	0.9625	0.9836	nan	nan	0.7355	0.9423	0.6157	0.8345	0.9387	0.9716	nan
0.0403	24.4444	220	0.1007	0.8397	0.9140	0.9705	nan	0.8826	0.9745	0.7393	0.9392	0.9633	0.9851	nan	nan	0.7353	0.9434	0.6172	0.8319	0.9388	0.9716	nan
0.0563	26.6667	240	0.1009	0.8406	0.9140	0.9704	nan	0.8914	0.9765	0.7306	0.9391	0.9603	0.9859	nan	nan	0.7360	0.9427	0.6202	0.8344	0.9383	0.9718	nan
0.0464	28.8889	260	0.1007	0.8404	0.9136	0.9704	nan	0.8876	0.9746	0.7358	0.9360	0.9631	0.9848	nan	nan	0.7360	0.9428	0.6178	0.8353	0.9386	0.9718	nan

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1