segformer-b0-finetuned-arabidopsis-roots-v02

This model is a fine-tuned version of nvidia/mit-b0 on the jacquelinegrimm/arabidopsis-roots-v02 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0500
• Mean Iou: 0.4449
• Mean Accuracy: 0.8899
• Overall Accuracy: 0.8899
• Accuracy Background: nan
• Accuracy Seedling: 0.8899
• Iou Background: 0.0
• Iou Seedling: 0.8899

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: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Seedling	Iou Background	Iou Seedling
0.4792	1.0	20	0.5681	0.4482	0.8964	0.8964	nan	0.8964	0.0	0.8964
0.3118	2.0	40	0.3829	0.4641	0.9282	0.9282	nan	0.9282	0.0	0.9282
0.2481	3.0	60	0.2563	0.4722	0.9445	0.9445	nan	0.9445	0.0	0.9445
0.2159	4.0	80	0.2402	0.4695	0.9389	0.9389	nan	0.9389	0.0	0.9389
0.1721	5.0	100	0.1644	0.4627	0.9254	0.9254	nan	0.9254	0.0	0.9254
0.1494	6.0	120	0.1407	0.4595	0.9189	0.9189	nan	0.9189	0.0	0.9189
0.1286	7.0	140	0.1213	0.4616	0.9233	0.9233	nan	0.9233	0.0	0.9233
0.1233	8.0	160	0.1097	0.4469	0.8939	0.8939	nan	0.8939	0.0	0.8939
0.108	9.0	180	0.0963	0.4419	0.8837	0.8837	nan	0.8837	0.0	0.8837
0.0968	10.0	200	0.0998	0.4491	0.8983	0.8983	nan	0.8983	0.0	0.8983
0.1041	11.0	220	0.0849	0.4354	0.8708	0.8708	nan	0.8708	0.0	0.8708
0.0842	12.0	240	0.0795	0.4444	0.8888	0.8888	nan	0.8888	0.0	0.8888
0.0896	13.0	260	0.0756	0.4402	0.8803	0.8803	nan	0.8803	0.0	0.8803
0.077	14.0	280	0.0797	0.4484	0.8968	0.8968	nan	0.8968	0.0	0.8968
0.0767	15.0	300	0.0690	0.4389	0.8779	0.8779	nan	0.8779	0.0	0.8779
0.0776	16.0	320	0.0665	0.4355	0.8710	0.8710	nan	0.8710	0.0	0.8710
0.0789	17.0	340	0.0680	0.4450	0.8901	0.8901	nan	0.8901	0.0	0.8901
0.071	18.0	360	0.0660	0.4424	0.8848	0.8848	nan	0.8848	0.0	0.8848
0.0661	19.0	380	0.0677	0.4371	0.8742	0.8742	nan	0.8742	0.0	0.8742
0.0784	20.0	400	0.0639	0.4402	0.8805	0.8805	nan	0.8805	0.0	0.8805
0.0789	21.0	420	0.0602	0.4369	0.8737	0.8737	nan	0.8737	0.0	0.8737
0.0756	22.0	440	0.0595	0.4313	0.8627	0.8627	nan	0.8627	0.0	0.8627
0.0563	23.0	460	0.0581	0.4420	0.8839	0.8839	nan	0.8839	0.0	0.8839
0.0713	24.0	480	0.0570	0.4438	0.8875	0.8875	nan	0.8875	0.0	0.8875
0.0594	25.0	500	0.0579	0.4478	0.8955	0.8955	nan	0.8955	0.0	0.8955
0.0636	26.0	520	0.0582	0.4523	0.9046	0.9046	nan	0.9046	0.0	0.9046
0.053	27.0	540	0.0558	0.4431	0.8863	0.8863	nan	0.8863	0.0	0.8863
0.0688	28.0	560	0.0561	0.4470	0.8939	0.8939	nan	0.8939	0.0	0.8939
0.0564	29.0	580	0.0542	0.4442	0.8885	0.8885	nan	0.8885	0.0	0.8885
0.0671	30.0	600	0.0537	0.4445	0.8891	0.8891	nan	0.8891	0.0	0.8891
0.0699	31.0	620	0.0530	0.4397	0.8794	0.8794	nan	0.8794	0.0	0.8794
0.0648	32.0	640	0.0535	0.4423	0.8845	0.8845	nan	0.8845	0.0	0.8845
0.058	33.0	660	0.0535	0.4467	0.8934	0.8934	nan	0.8934	0.0	0.8934
0.0582	34.0	680	0.0536	0.4513	0.9025	0.9025	nan	0.9025	0.0	0.9025
0.0561	35.0	700	0.0525	0.4444	0.8888	0.8888	nan	0.8888	0.0	0.8888
0.0731	36.0	720	0.0520	0.4439	0.8878	0.8878	nan	0.8878	0.0	0.8878
0.0503	37.0	740	0.0528	0.4484	0.8968	0.8968	nan	0.8968	0.0	0.8968
0.0895	38.0	760	0.0505	0.4388	0.8777	0.8777	nan	0.8777	0.0	0.8777
0.064	39.0	780	0.0508	0.4418	0.8836	0.8836	nan	0.8836	0.0	0.8836
0.0481	40.0	800	0.0512	0.4474	0.8949	0.8949	nan	0.8949	0.0	0.8949
0.0497	41.0	820	0.0533	0.4503	0.9006	0.9006	nan	0.9006	0.0	0.9006
0.06	42.0	840	0.0511	0.4447	0.8893	0.8893	nan	0.8893	0.0	0.8893
0.0543	43.0	860	0.0504	0.4441	0.8882	0.8882	nan	0.8882	0.0	0.8882
0.0539	44.0	880	0.0515	0.4421	0.8843	0.8843	nan	0.8843	0.0	0.8843
0.0461	45.0	900	0.0506	0.4461	0.8921	0.8921	nan	0.8921	0.0	0.8921
0.0554	46.0	920	0.0501	0.4453	0.8907	0.8907	nan	0.8907	0.0	0.8907
0.0475	47.0	940	0.0503	0.4471	0.8942	0.8942	nan	0.8942	0.0	0.8942
0.0645	48.0	960	0.0506	0.4463	0.8926	0.8926	nan	0.8926	0.0	0.8926
0.045	49.0	980	0.0510	0.4471	0.8942	0.8942	nan	0.8942	0.0	0.8942
0.0576	50.0	1000	0.0500	0.4449	0.8899	0.8899	nan	0.8899	0.0	0.8899

Framework versions

• Transformers 4.36.2
• Pytorch 2.0.0
• Datasets 2.1.0
• Tokenizers 0.15.0