segformer-b5-cityscapes-finetuned-coastTrain-grCoastline

This model is a fine-tuned version of peldrak/segformer-b5-cityscapes-finetuned-coastTrain on the peldrak/grCoastline_512 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2369
• Mean Iou: 0.7581
• Mean Accuracy: 0.8319
• Overall Accuracy: 0.9386
• Accuracy Water: 0.9821
• Accuracy Whitewater: 0.4902
• Accuracy Sediment: 0.9136
• Accuracy Other Natural Terrain: 0.8026
• Accuracy Vegetation: 0.9297
• Accuracy Development: 0.7063
• Accuracy Unknown: 0.9989
• Iou Water: 0.9520
• Iou Whitewater: 0.3507
• Iou Sediment: 0.8649
• Iou Other Natural Terrain: 0.6679
• Iou Vegetation: 0.8298
• Iou Development: 0.6457
• Iou Unknown: 0.9956
• F1 Score: 0.9379

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: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Water	Accuracy Whitewater	Accuracy Sediment	Accuracy Other Natural Terrain	Accuracy Vegetation	Accuracy Development	Accuracy Unknown	Iou Water	Iou Whitewater	Iou Sediment	Iou Other Natural Terrain	Iou Vegetation	Iou Development	Iou Unknown	F1 Score
0.4436	0.24	20	0.3559	0.6026	0.6838	0.8880	0.9773	0.0337	0.9374	0.5772	0.8436	0.4204	0.9972	0.8777	0.0293	0.7012	0.4913	0.7446	0.3805	0.9937	0.8816
0.3897	0.49	40	0.2428	0.6524	0.7222	0.9165	0.9825	0.0	0.9378	0.7357	0.8910	0.5144	0.9944	0.9418	0.0	0.7760	0.5956	0.8037	0.4564	0.9931	0.9134
0.5041	0.73	60	0.2645	0.6562	0.7196	0.9172	0.9864	0.0587	0.9317	0.6634	0.9456	0.4577	0.9934	0.9407	0.0581	0.7808	0.5880	0.8037	0.4299	0.9922	0.9118
0.1355	0.98	80	0.1992	0.7084	0.7729	0.9324	0.9855	0.1154	0.9117	0.8215	0.8980	0.6808	0.9974	0.9470	0.1086	0.8482	0.6539	0.8245	0.5821	0.9943	0.9316
0.1142	1.22	100	0.2597	0.6630	0.7299	0.9220	0.9852	0.0279	0.9498	0.8028	0.8860	0.4611	0.9965	0.9489	0.0278	0.7668	0.6533	0.8202	0.4298	0.9943	0.9184
0.2346	1.46	120	0.2670	0.6708	0.7331	0.9239	0.9823	0.0530	0.9331	0.7716	0.9238	0.4729	0.9953	0.9497	0.0524	0.8044	0.6352	0.8128	0.4478	0.9936	0.9201
0.4209	1.71	140	0.1952	0.7291	0.7955	0.9336	0.9688	0.2746	0.9536	0.8644	0.8830	0.6252	0.9989	0.9485	0.2476	0.8375	0.6754	0.8291	0.5719	0.9939	0.9328
0.0803	1.95	160	0.2460	0.6772	0.7440	0.9228	0.9824	0.0952	0.9514	0.7835	0.8898	0.5101	0.9958	0.9458	0.0937	0.7832	0.6355	0.8190	0.4692	0.9943	0.9200
0.3305	2.2	180	0.2127	0.7372	0.7992	0.9381	0.9869	0.1967	0.9330	0.7088	0.9337	0.8417	0.9935	0.9463	0.1844	0.8578	0.6514	0.8284	0.6996	0.9928	0.9366
0.1355	2.44	200	0.1968	0.7219	0.7768	0.9387	0.9877	0.0983	0.9309	0.7413	0.9411	0.7414	0.9971	0.9480	0.0963	0.8611	0.6629	0.8333	0.6575	0.9941	0.9370
0.0807	2.68	220	0.2531	0.6939	0.7569	0.9273	0.9826	0.1709	0.9534	0.7590	0.9194	0.5160	0.9969	0.9520	0.1628	0.8076	0.6326	0.8290	0.4785	0.9950	0.9242
0.1226	2.93	240	0.2434	0.7382	0.8081	0.9315	0.9832	0.3932	0.9090	0.7441	0.9272	0.7026	0.9976	0.9340	0.3119	0.8604	0.6199	0.8192	0.6276	0.9947	0.9303
0.0681	3.17	260	0.2265	0.7529	0.8220	0.9387	0.9802	0.3786	0.9371	0.7742	0.9194	0.7668	0.9974	0.9506	0.3042	0.8661	0.6614	0.8315	0.6617	0.9950	0.9380
0.1304	3.41	280	0.2342	0.7360	0.8009	0.9348	0.9772	0.3403	0.9507	0.7835	0.9220	0.6358	0.9971	0.9486	0.2864	0.8471	0.6572	0.8302	0.5880	0.9947	0.9333
0.1353	3.66	300	0.1970	0.7426	0.8071	0.9400	0.9803	0.2770	0.9285	0.8866	0.8962	0.6822	0.9987	0.9508	0.2259	0.8732	0.6883	0.8370	0.6278	0.9954	0.9397
0.3968	3.9	320	0.2181	0.7551	0.8214	0.9389	0.9868	0.3918	0.9280	0.7945	0.9160	0.7343	0.9982	0.9512	0.3204	0.8645	0.6646	0.8297	0.6594	0.9962	0.9381
0.0548	4.15	340	0.2025	0.7645	0.8313	0.9411	0.9804	0.4091	0.9287	0.7892	0.9208	0.7920	0.9987	0.9525	0.3393	0.8686	0.6729	0.8347	0.6878	0.9956	0.9406
0.0653	4.39	360	0.2549	0.7399	0.8105	0.9327	0.9851	0.4396	0.9222	0.8119	0.9144	0.6029	0.9976	0.9528	0.3534	0.8455	0.6479	0.8239	0.5602	0.9957	0.9314
0.1244	4.63	380	0.2580	0.7216	0.7821	0.9336	0.9925	0.2673	0.9136	0.7674	0.9405	0.5951	0.9984	0.9465	0.2327	0.8327	0.6598	0.8338	0.5501	0.9957	0.9312
0.2298	4.88	400	0.2345	0.7441	0.8112	0.9368	0.9760	0.2900	0.9511	0.8680	0.8716	0.7250	0.9965	0.9520	0.2585	0.8639	0.6659	0.8229	0.6497	0.9956	0.9369
0.0469	5.12	420	0.2614	0.7131	0.7813	0.9279	0.9854	0.3576	0.9420	0.8114	0.9099	0.4639	0.9986	0.9534	0.3075	0.8081	0.6597	0.8237	0.4432	0.9957	0.9244
0.1398	5.37	440	0.2542	0.7344	0.8077	0.9318	0.9851	0.4151	0.9388	0.8538	0.8842	0.5789	0.9979	0.9561	0.3381	0.8258	0.6665	0.8241	0.5345	0.9957	0.9305
0.0683	5.61	460	0.2496	0.7472	0.8213	0.9365	0.9856	0.5096	0.9419	0.8014	0.9301	0.5823	0.9980	0.9569	0.3788	0.8228	0.6847	0.8445	0.5467	0.9961	0.9346
0.141	5.85	480	0.2251	0.7514	0.8174	0.9393	0.9873	0.3664	0.9389	0.8465	0.8977	0.6867	0.9982	0.9559	0.3151	0.8544	0.6856	0.8387	0.6140	0.9962	0.9387
0.2161	6.1	500	0.2369	0.7581	0.8319	0.9386	0.9821	0.4902	0.9136	0.8026	0.9297	0.7063	0.9989	0.9520	0.3507	0.8649	0.6679	0.8298	0.6457	0.9956	0.9379

Framework versions

• Transformers 4.38.1
• Pytorch 2.1.2
• Datasets 2.18.0
• Tokenizers 0.15.2