segformer-b5-finetuned-apple-dms-run4

This model is a fine-tuned version of nvidia/segformer-b5-finetuned-ade-640-640 on the AllanK24/apple-dms-materials dataset. It achieves the following results on the evaluation set:

Loss: 9.7669
Mean Iou: 0.3836
Mean Accuracy: 0.4480
Overall Accuracy: 0.7177
Iou No Label: 0.0
Accuracy Animal Skin: 0.5329
Iou Animal Skin: 0.3352
Accuracy Bone Teeth Horn: 0.0
Iou Bone Teeth Horn: 0.0
Accuracy Brickwork: 0.6398
Iou Brickwork: 0.5657
Accuracy Cardboard: 0.4971
Iou Cardboard: 0.4327
Accuracy Carpet Rug: 0.7372
Iou Carpet Rug: 0.6823
Accuracy Ceiling Tile: 0.7805
Iou Ceiling Tile: 0.7238
Accuracy Ceramic: 0.6910
Iou Ceramic: 0.5952
Accuracy Chalkboard Blackboard: 0.6698
Iou Chalkboard Blackboard: 0.6127
Accuracy Clutter: 0.0
Iou Clutter: 0.0
Accuracy Concrete: 0.4339
Iou Concrete: 0.3413
Accuracy Cork Corkboard: 0.0
Iou Cork Corkboard: 0.0
Accuracy Engineered Stone: 0.0
Iou Engineered Stone: 0.0
Accuracy Fabric Cloth: 0.8402
Iou Fabric Cloth: 0.7735
Accuracy Fiberglass Wool: 0.0
Iou Fiberglass Wool: 0.0
Accuracy Fire: 0.0
Iou Fire: 0.0
Accuracy Foliage: 0.9046
Iou Foliage: 0.8234
Accuracy Food: 0.8635
Iou Food: 0.7769
Accuracy Fur: 0.8762
Iou Fur: 0.7877
Accuracy Gemstone Quartz: 0.0
Iou Gemstone Quartz: 0.0
Accuracy Glass: 0.6544
Iou Glass: 0.5740
Accuracy Hair: 0.8260
Iou Hair: 0.7179
Accuracy Ice: 0.0000
Iou Ice: 0.0000
Accuracy Leather: 0.4178
Iou Leather: 0.3721
Accuracy Liquid Non-water: 0.0
Iou Liquid Non-water: 0.0
Accuracy Metal: 0.3613
Iou Metal: 0.3048
Accuracy Mirror: 0.4897
Iou Mirror: 0.4201
Accuracy Paint Plaster Enamel: 0.7484
Iou Paint Plaster Enamel: 0.6911
Accuracy Paper: 0.6480
Iou Paper: 0.5575
Accuracy Pearl: 0.0
Iou Pearl: 0.0
Accuracy Photograph Painting: 0.3260
Iou Photograph Painting: 0.2795
Accuracy Plastic Clear: 0.2786
Iou Plastic Clear: 0.2363
Accuracy Plastic Non-clear: 0.3583
Iou Plastic Non-clear: 0.3052
Accuracy Rubber Latex: 0.2241
Iou Rubber Latex: 0.2069
Accuracy Sand: 0.6068
Iou Sand: 0.5008
Accuracy Skin Lips: 0.8279
Iou Skin Lips: 0.7208
Accuracy Sky: 0.9608
Iou Sky: 0.9264
Accuracy Snow: 0.7195
Iou Snow: 0.5864
Accuracy Soap: 0.0
Iou Soap: 0.0
Accuracy Soil Mud: 0.5570
Iou Soil Mud: 0.4064
Accuracy Sponge: 0.0
Iou Sponge: 0.0
Accuracy Stone Natural: 0.6388
Iou Stone Natural: 0.5136
Accuracy Stone Polished: 0.1924
Iou Stone Polished: 0.1702
Accuracy Styrofoam: 0.0
Iou Styrofoam: 0.0
Accuracy Tile: 0.6683
Iou Tile: 0.6118
Accuracy Wallpaper: 0.5190
Iou Wallpaper: 0.4231
Accuracy Water: 0.8852
Iou Water: 0.8063
Accuracy Wax: 0.0
Iou Wax: 0.0
Accuracy Whiteboard: 0.7443
Iou Whiteboard: 0.6706
Accuracy Wicker: 0.4696
Iou Wicker: 0.4252
Accuracy Wood: 0.7253
Iou Wood: 0.6696
Accuracy Wood Tree: 0.4098
Iou Wood Tree: 0.3165
Accuracy Asphalt: 0.5724
Iou Asphalt: 0.4690

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: 0.0001
train_batch_size: 32
eval_batch_size: 16
seed: 42
distributed_type: multi-GPU
num_devices: 8
total_train_batch_size: 256
total_eval_batch_size: 128
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 0.1
num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Animal Skin	Iou Animal Skin	Accuracy Brickwork	Iou Brickwork	Accuracy Cardboard	Iou Cardboard	Accuracy Carpet Rug	Iou Carpet Rug	Accuracy Ceiling Tile	Iou Ceiling Tile	Accuracy Ceramic	Iou Ceramic	Accuracy Chalkboard Blackboard	Iou Chalkboard Blackboard	Accuracy Concrete	Iou Concrete	Accuracy Fabric Cloth	Iou Fabric Cloth	Accuracy Foliage	Iou Foliage	Accuracy Food	Iou Food	Accuracy Fur	Iou Fur	Accuracy Glass	Iou Glass	Accuracy Hair	Iou Hair	Accuracy Ice	Iou Ice	Accuracy Leather	Iou Leather	Accuracy Metal	Iou Metal	Accuracy Mirror	Iou Mirror	Accuracy Paint Plaster Enamel	Iou Paint Plaster Enamel	Accuracy Paper	Iou Paper	Accuracy Photograph Painting	Iou Photograph Painting	Accuracy Plastic Clear	Iou Plastic Clear	Accuracy Plastic Non-clear	Iou Plastic Non-clear	Accuracy Rubber Latex	Iou Rubber Latex	Accuracy Sand	Iou Sand	Accuracy Skin Lips	Iou Skin Lips	Accuracy Sky	Iou Sky	Accuracy Snow	Iou Snow	Accuracy Soil Mud	Iou Soil Mud	Accuracy Stone Natural	Iou Stone Natural	Accuracy Stone Polished	Iou Stone Polished	Accuracy Tile	Iou Tile	Accuracy Wallpaper	Iou Wallpaper	Accuracy Water	Iou Water	Accuracy Whiteboard	Iou Whiteboard	Accuracy Wicker	Iou Wicker	Accuracy Wood	Iou Wood	Accuracy Wood Tree	Iou Wood Tree	Accuracy Asphalt	Iou Asphalt
25.1467	1.7045	150	16.2675	0.1132	0.1514	0.6230	0.0	0.0	0.0	0.0	0.0	0.0	0.7795	0.5593	0.5215	0.5078	0.0	0.0	0.0	0.0	0.0	0.0	0.8860	0.6011	0.9265	0.5751	0.0	0.0	0.0000	0.0000	0.5776	0.4325	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.8117	0.6603	0.1450	0.1379	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0128	0.0127	0.9521	0.8845	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.6299	0.5287	0.0	0.0	0.7965	0.4936	0.0	0.0	0.0	0.0	0.8358	0.6044	0.0	0.0	0.0	0.0
11.5476	3.4091	300	9.6860	0.2209	0.2648	0.6910	0.0	0.0	0.1197	0.1190	0.0	0.0	0.8308	0.7160	0.8796	0.7577	0.4310	0.3981	0.0	0.0	0.1160	0.1071	0.8445	0.7592	0.9260	0.7892	0.8810	0.7002	0.8588	0.7452	0.6745	0.5411	0.7625	0.6340	0.0	0.0	0.0	0.0	0.1785	0.1682	0.0	0.0	0.8043	0.7023	0.6639	0.5187	0.0	0.0	0.0	0.0	0.1435	0.1383	0.0	0.0	0.0	0.0	0.7474	0.6239	0.9561	0.9120	0.0	0.0	0.0	0.0	0.6112	0.3062	0.0	0.0	0.7058	0.6278	0.0001	0.0001	0.8831	0.7669	0.0	0.0	0.0	0.0	0.7488	0.6776	0.0	0.0	0.0	0.0
7.7487	5.1136	450	8.6804	0.2867	0.3361	0.7020	0.0	0.0	0.6977	0.5349	0.0491	0.0490	0.7391	0.6776	0.8778	0.7673	0.6555	0.5518	0.0	0.0	0.4109	0.3148	0.8158	0.7530	0.9111	0.8125	0.8536	0.7624	0.8722	0.7665	0.6503	0.5635	0.7601	0.6632	0.0	0.0	0.4308	0.3936	0.2965	0.2604	0.4482	0.3819	0.7677	0.6948	0.6477	0.5447	0.1359	0.1258	0.0	0.0	0.3433	0.2900	0.0	0.0	0.0	0.0	0.7623	0.6649	0.9712	0.9231	0.0	0.0	0.4797	0.3205	0.6433	0.4399	0.0	0.0	0.6916	0.6292	0.5776	0.4568	0.8744	0.8016	0.2953	0.2849	0.0	0.0	0.7132	0.6602	0.0	0.0	0.1061	0.1046
6.2739	6.8182	600	8.5191	0.3443	0.4027	0.7262	0.0001	0.0001	0.6555	0.5683	0.4437	0.3967	0.7805	0.7079	0.7594	0.7183	0.6825	0.5844	0.5905	0.5635	0.4883	0.3552	0.8383	0.7660	0.8966	0.8111	0.8367	0.7551	0.8869	0.7539	0.6530	0.5614	0.8048	0.6939	0.0	0.0	0.4905	0.4386	0.3738	0.3079	0.4832	0.4054	0.7710	0.7005	0.6503	0.5474	0.2757	0.2422	0.1504	0.1421	0.3705	0.3075	0.0	0.0	0.0472	0.0472	0.7914	0.6861	0.9673	0.9229	0.4545	0.4167	0.6243	0.3636	0.5801	0.4860	0.0594	0.0590	0.7319	0.6457	0.6155	0.4579	0.8629	0.7931	0.6446	0.6077	0.2505	0.2444	0.7703	0.6966	0.0574	0.0571	0.6017	0.4380
5.3600	8.5227	750	8.6966	0.3690	0.4332	0.7191	0.0990	0.0865	0.6982	0.5663	0.5323	0.4480	0.6986	0.6544	0.8258	0.7567	0.7309	0.6070	0.6778	0.6263	0.5353	0.3653	0.8467	0.7747	0.8960	0.8216	0.8397	0.7644	0.8886	0.7739	0.6373	0.5597	0.8127	0.6984	0.0	0.0	0.4283	0.3969	0.4297	0.3304	0.5558	0.4088	0.7380	0.6856	0.6440	0.5535	0.3568	0.2941	0.2768	0.2358	0.3699	0.3112	0.0314	0.0314	0.3115	0.2864	0.8147	0.7002	0.9614	0.9221	0.6418	0.5581	0.4801	0.3584	0.5878	0.4726	0.2118	0.1863	0.6967	0.6229	0.5684	0.4473	0.8880	0.7986	0.7656	0.6850	0.4406	0.4015	0.7542	0.6860	0.3281	0.2646	0.5286	0.4183
4.7475	10.2273	900	9.1646	0.3711	0.4275	0.7084	0.2707	0.1946	0.5871	0.5353	0.4686	0.4171	0.7955	0.7233	0.8186	0.7511	0.7026	0.5975	0.6779	0.6220	0.4149	0.3199	0.8359	0.7701	0.8978	0.8134	0.8649	0.7740	0.8833	0.7750	0.6366	0.5653	0.8161	0.7072	0.0	0.0	0.3628	0.3356	0.3289	0.2849	0.4913	0.4144	0.7407	0.6839	0.6322	0.5465	0.2536	0.2323	0.2390	0.2085	0.3090	0.2720	0.1732	0.1680	0.4216	0.3801	0.8153	0.7085	0.9556	0.9202	0.6839	0.5803	0.5360	0.4003	0.5941	0.5011	0.1509	0.1453	0.6806	0.6204	0.5072	0.4190	0.8743	0.7969	0.7194	0.6519	0.4346	0.4018	0.7232	0.6664	0.3825	0.3015	0.5513	0.4611
4.3228	11.9318	1050	9.1586	0.3826	0.4458	0.7248	0.4363	0.2874	0.6269	0.5634	0.4522	0.4001	0.7400	0.6806	0.8163	0.7516	0.7077	0.6021	0.6446	0.6083	0.4335	0.3360	0.8430	0.7743	0.8994	0.8193	0.8772	0.7831	0.8759	0.7895	0.6461	0.5725	0.8231	0.7116	0.0	0.0	0.4889	0.4264	0.3765	0.3167	0.5331	0.4232	0.7587	0.6967	0.6789	0.5625	0.3065	0.2677	0.2468	0.2176	0.3710	0.3155	0.2098	0.1965	0.5815	0.4720	0.8294	0.7118	0.9598	0.9273	0.7105	0.5960	0.5665	0.4073	0.6107	0.5092	0.2175	0.1811	0.6786	0.6160	0.4811	0.4032	0.8783	0.8050	0.7369	0.6676	0.4808	0.4316	0.7505	0.6830	0.3832	0.3030	0.5223	0.4619
4.0235	13.6364	1200	9.4911	0.3820	0.4453	0.7232	0.4486	0.3079	0.6248	0.5579	0.4550	0.4074	0.7417	0.6830	0.8054	0.7401	0.6786	0.5880	0.6733	0.6149	0.4383	0.3396	0.8406	0.7722	0.9087	0.8234	0.8645	0.7753	0.8560	0.7805	0.6894	0.5903	0.8285	0.7156	0.0	0.0	0.4220	0.3761	0.3531	0.3005	0.5140	0.4299	0.7478	0.6914	0.6487	0.5545	0.3319	0.2827	0.2714	0.2320	0.3614	0.3078	0.2111	0.1979	0.5664	0.4725	0.8253	0.7157	0.9616	0.9264	0.6924	0.5855	0.5533	0.4002	0.6150	0.5015	0.1839	0.1662	0.7462	0.6469	0.5197	0.4201	0.8821	0.8035	0.7385	0.6654	0.4734	0.4282	0.7341	0.6732	0.3978	0.3112	0.5487	0.4609
3.8194	15.3409	1350	9.6066	0.3843	0.4512	0.7232	0.5448	0.3228	0.6467	0.5708	0.5022	0.4292	0.7516	0.6934	0.7954	0.7340	0.6947	0.5974	0.6654	0.6134	0.4457	0.3452	0.8408	0.7743	0.9054	0.8233	0.8646	0.7763	0.8808	0.7909	0.6396	0.5674	0.8311	0.7176	0.0	0.0	0.3971	0.3590	0.3634	0.3052	0.4825	0.4166	0.7563	0.6950	0.6593	0.5641	0.3487	0.2930	0.2872	0.2423	0.4090	0.3308	0.2240	0.2063	0.5696	0.4882	0.8333	0.7188	0.9602	0.9267	0.7171	0.5826	0.5956	0.4094	0.6344	0.5167	0.1679	0.1543	0.6737	0.6143	0.5193	0.4255	0.8838	0.8068	0.7110	0.6559	0.4827	0.4324	0.7327	0.6731	0.4070	0.3168	0.6362	0.4761
3.6946	17.0455	1500	9.7223	0.3834	0.4479	0.7189	0.5307	0.3427	0.6388	0.5667	0.5012	0.4354	0.7584	0.6963	0.7756	0.7212	0.6977	0.5985	0.6648	0.6088	0.4384	0.3427	0.8311	0.7712	0.9068	0.8236	0.8654	0.7770	0.8728	0.7880	0.6540	0.5746	0.8233	0.7178	0.0000	0.0000	0.4134	0.3700	0.3584	0.3038	0.4929	0.4209	0.7530	0.6934	0.6474	0.5574	0.3203	0.2763	0.2724	0.2328	0.3590	0.3053	0.2191	0.2036	0.6208	0.5057	0.8324	0.7206	0.9615	0.9277	0.7217	0.5880	0.5634	0.4051	0.6307	0.5117	0.1868	0.1647	0.6707	0.6139	0.4939	0.4109	0.8863	0.8060	0.7283	0.6619	0.4559	0.4164	0.7298	0.6707	0.4021	0.3135	0.6097	0.4749
3.6308	18.75	1650	9.7669	0.3836	0.4480	0.7177	0.5329	0.3352	0.6398	0.5657	0.4971	0.4327	0.7372	0.6823	0.7805	0.7238	0.6910	0.5952	0.6698	0.6127	0.4339	0.3413	0.8402	0.7735	0.9046	0.8234	0.8635	0.7769	0.8762	0.7877	0.6544	0.5740	0.8260	0.7179	0.0000	0.0000	0.4178	0.3721	0.3613	0.3048	0.4897	0.4201	0.7484	0.6911	0.6480	0.5575	0.3260	0.2795	0.2786	0.2363	0.3583	0.3052	0.2241	0.2069	0.6068	0.5008	0.8279	0.7208	0.9608	0.9264	0.7195	0.5864	0.5570	0.4064	0.6388	0.5136	0.1924	0.1702	0.6683	0.6118	0.5190	0.4231	0.8852	0.8063	0.7443	0.6706	0.4696	0.4252	0.7253	0.6696	0.4098	0.3165	0.5724	0.4690

Framework versions

Transformers 5.0.0
Pytorch 2.9.1+cu128
Datasets 4.5.0
Tokenizers 0.22.2

Downloads last month: 9

Safetensors

Model size

84.6M params

Tensor type

F32

Model tree for AllanK24/segformer-b5-finetuned-apple-dms-run4

Base model

nvidia/segformer-b5-finetuned-ade-640-640

Finetuned

(22)

this model