segformer-b5-finetuned-apple-dms-run10

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: 1.7623
• Mean Iou: 0.4532
• Mean Accuracy: 0.5406
• Overall Accuracy: 0.8224
• Accuracy Animal Skin: 0.6755
• Iou Animal Skin: 0.4518
• Accuracy Bone Teeth Horn: 0.0
• Iou Bone Teeth Horn: 0.0
• Accuracy Brickwork: 0.7190
• Iou Brickwork: 0.6032
• Accuracy Cardboard: 0.6659
• Iou Cardboard: 0.5084
• Accuracy Carpet Rug: 0.8853
• Iou Carpet Rug: 0.7646
• Accuracy Ceiling Tile: 0.8605
• Iou Ceiling Tile: 0.7578
• Accuracy Ceramic: 0.7609
• Iou Ceramic: 0.6274
• Accuracy Chalkboard Blackboard: 0.7714
• Iou Chalkboard Blackboard: 0.6618
• Accuracy Clutter: 0.0209
• Iou Clutter: 0.0189
• Accuracy Concrete: 0.6176
• Iou Concrete: 0.4249
• Accuracy Cork Corkboard: 0.0777
• Iou Cork Corkboard: 0.0771
• Accuracy Engineered Stone: 0.0887
• Iou Engineered Stone: 0.0831
• Accuracy Fabric Cloth: 0.9012
• Iou Fabric Cloth: 0.8021
• Accuracy Fiberglass Wool: 0.0
• Iou Fiberglass Wool: 0.0
• Accuracy Fire: 0.3660
• Iou Fire: 0.3238
• Accuracy Foliage: 0.9401
• Iou Foliage: 0.8457
• Accuracy Food: 0.9134
• Iou Food: 0.7870
• Accuracy Fur: 0.9191
• Iou Fur: 0.8267
• Accuracy Gemstone Quartz: 0.0206
• Iou Gemstone Quartz: 0.0165
• Accuracy Glass: 0.7480
• Iou Glass: 0.6279
• Accuracy Hair: 0.8541
• Iou Hair: 0.7449
• Accuracy Ice: 0.2601
• Iou Ice: 0.2271
• Accuracy Leather: 0.6014
• Iou Leather: 0.5052
• Accuracy Liquid Non-water: 0.1274
• Iou Liquid Non-water: 0.1127
• Accuracy Metal: 0.4793
• Iou Metal: 0.3624
• Accuracy Mirror: 0.5963
• Iou Mirror: 0.5059
• Accuracy Paint Plaster Enamel: 0.8822
• Iou Paint Plaster Enamel: 0.7548
• Accuracy Paper: 0.7413
• Iou Paper: 0.6046
• Accuracy Pearl: 0.0
• Iou Pearl: 0.0
• Accuracy Photograph Painting: 0.4709
• Iou Photograph Painting: 0.3654
• Accuracy Plastic Clear: 0.3889
• Iou Plastic Clear: 0.2853
• Accuracy Plastic Non-clear: 0.5811
• Iou Plastic Non-clear: 0.4103
• Accuracy Rubber Latex: 0.2984
• Iou Rubber Latex: 0.2706
• Accuracy Sand: 0.6261
• Iou Sand: 0.4917
• Accuracy Skin Lips: 0.8534
• Iou Skin Lips: 0.7521
• Accuracy Sky: 0.9745
• Iou Sky: 0.9388
• Accuracy Snow: 0.7361
• Iou Snow: 0.6369
• Accuracy Soap: 0.0
• Iou Soap: 0.0
• Accuracy Soil Mud: 0.5831
• Iou Soil Mud: 0.4231
• Accuracy Sponge: 0.0
• Iou Sponge: 0.0
• Accuracy Stone Natural: 0.6938
• Iou Stone Natural: 0.5581
• Accuracy Stone Polished: 0.3082
• Iou Stone Polished: 0.2441
• Accuracy Styrofoam: 0.0
• Iou Styrofoam: 0.0
• Accuracy Tile: 0.8167
• Iou Tile: 0.6844
• Accuracy Wallpaper: 0.5244
• Iou Wallpaper: 0.4070
• Accuracy Water: 0.9011
• Iou Water: 0.8107
• Accuracy Wax: 0.4492
• Iou Wax: 0.4449
• Accuracy Whiteboard: 0.8354
• Iou Whiteboard: 0.7040
• Accuracy Wicker: 0.5684
• Iou Wicker: 0.4837
• Accuracy Wood: 0.8578
• Iou Wood: 0.7432
• Accuracy Wood Tree: 0.5182
• Iou Wood Tree: 0.3877
• Accuracy Asphalt: 0.6330
• Iou Asphalt: 0.4985

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: 0.001
• 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
• label_smoothing_factor: 0.2

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Animal Skin	Iou Animal Skin	Accuracy Bone Teeth Horn	Iou Bone Teeth Horn	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 Clutter	Iou Clutter	Accuracy Concrete	Iou Concrete	Accuracy Cork Corkboard	Iou Cork Corkboard	Accuracy Engineered Stone	Iou Engineered Stone	Accuracy Fabric Cloth	Iou Fabric Cloth	Accuracy Fiberglass Wool	Iou Fiberglass Wool	Accuracy Fire	Iou Fire	Accuracy Foliage	Iou Foliage	Accuracy Food	Iou Food	Accuracy Fur	Iou Fur	Accuracy Gemstone Quartz	Iou Gemstone Quartz	Accuracy Glass	Iou Glass	Accuracy Hair	Iou Hair	Accuracy Ice	Iou Ice	Accuracy Leather	Iou Leather	Accuracy Liquid Non-water	Iou Liquid Non-water	Accuracy Metal	Iou Metal	Accuracy Mirror	Iou Mirror	Accuracy Paint Plaster Enamel	Iou Paint Plaster Enamel	Accuracy Paper	Iou Paper	Accuracy Pearl	Iou Pearl	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 Soap	Iou Soap	Accuracy Soil Mud	Iou Soil Mud	Accuracy Sponge	Iou Sponge	Accuracy Stone Natural	Iou Stone Natural	Accuracy Stone Polished	Iou Stone Polished	Accuracy Styrofoam	Iou Styrofoam	Accuracy Tile	Iou Tile	Accuracy Wallpaper	Iou Wallpaper	Accuracy Water	Iou Water	Accuracy Wax	Iou Wax	Accuracy Whiteboard	Iou Whiteboard	Accuracy Wicker	Iou Wicker	Accuracy Wood	Iou Wood	Accuracy Wood Tree	Iou Wood Tree	Accuracy Asphalt	Iou Asphalt
2.6193	1.7045	150	1.8642	0.2821	0.3476	0.7673	0.0	0.0	0.0	0.0	0.6057	0.4818	0.3378	0.2951	0.8666	0.6817	0.7559	0.7038	0.6468	0.5399	0.0	0.0	0.0	0.0	0.5856	0.3512	0.0	0.0	0.0	0.0	0.8997	0.7296	0.0	0.0	0.0	0.0	0.9295	0.8081	0.8663	0.7632	0.9250	0.7312	0.0	0.0	0.7066	0.5480	0.7776	0.6620	0.0	0.0	0.1158	0.1113	0.0	0.0	0.2884	0.2353	0.4135	0.3635	0.8811	0.7023	0.7941	0.4967	0.0	0.0	0.3673	0.2405	0.0189	0.0188	0.3949	0.2972	0.0	0.0	0.0045	0.0045	0.7649	0.6506	0.9660	0.9170	0.0	0.0	0.0	0.0	0.5459	0.2846	0.0	0.0	0.6543	0.4633	0.0439	0.0438	0.0	0.0	0.6978	0.6032	0.1110	0.1078	0.9223	0.7673	0.0	0.0	0.1463	0.1418	0.0001	0.0001	0.7926	0.6851	0.0031	0.0031	0.2456	0.2381
1.7355	3.4091	300	1.7668	0.4046	0.4972	0.8030	0.5913	0.4254	0.0	0.0	0.7720	0.5484	0.6033	0.4574	0.9162	0.6958	0.8729	0.7558	0.7240	0.5942	0.7488	0.6230	0.0	0.0	0.6174	0.3871	0.0	0.0	0.0	0.0	0.8968	0.7796	0.0	0.0	0.0	0.0	0.9396	0.8300	0.8924	0.7827	0.9300	0.8202	0.0	0.0	0.6984	0.5928	0.8213	0.7092	0.0002	0.0002	0.5808	0.4909	0.0	0.0	0.4542	0.3241	0.5424	0.4592	0.8673	0.7334	0.7827	0.5697	0.0	0.0	0.5307	0.3418	0.3438	0.2548	0.4151	0.3378	0.2082	0.1953	0.5337	0.4660	0.8315	0.7113	0.9733	0.9313	0.6864	0.6194	0.0	0.0	0.5607	0.3887	0.0	0.0	0.7327	0.4994	0.2578	0.2195	0.0	0.0	0.7267	0.6399	0.6744	0.4793	0.9066	0.7942	0.0	0.0	0.7937	0.6195	0.4981	0.4386	0.8444	0.7199	0.4196	0.3075	0.6639	0.4970
1.6191	5.1136	450	1.7432	0.4243	0.5097	0.8144	0.6092	0.4588	0.0	0.0	0.7118	0.5979	0.6956	0.4453	0.8702	0.7497	0.8831	0.7652	0.7527	0.6141	0.7189	0.6294	0.0	0.0	0.5582	0.3853	0.0	0.0	0.0	0.0	0.8932	0.7891	0.0	0.0	0.1717	0.1618	0.9278	0.8388	0.9053	0.7878	0.9360	0.8334	0.0	0.0	0.7627	0.6177	0.8198	0.7222	0.1795	0.1531	0.5574	0.4845	0.0	0.0	0.4838	0.3552	0.5726	0.4888	0.8773	0.7464	0.7667	0.5842	0.0	0.0	0.4415	0.3421	0.3711	0.2815	0.5224	0.3888	0.2410	0.2261	0.6382	0.5091	0.8460	0.7301	0.9683	0.9326	0.7487	0.6580	0.0	0.0	0.6673	0.4500	0.0	0.0	0.6798	0.5481	0.2790	0.2380	0.0	0.0	0.8388	0.6661	0.5316	0.4360	0.9055	0.8073	0.0	0.0	0.7628	0.6489	0.5316	0.4575	0.8372	0.7294	0.4348	0.3176	0.6057	0.4857
1.5588	6.8182	600	1.7473	0.4308	0.5269	0.8149	0.6117	0.4138	0.0	0.0	0.7434	0.5984	0.6801	0.4788	0.8850	0.7525	0.8566	0.7600	0.7356	0.6199	0.7765	0.6598	0.0	0.0	0.6434	0.4104	0.0	0.0	0.0	0.0	0.8960	0.7961	0.0	0.0	0.4068	0.2696	0.9267	0.8439	0.9022	0.7866	0.9470	0.8377	0.0024	0.0023	0.7816	0.6154	0.8516	0.7305	0.2744	0.2226	0.6143	0.5076	0.0	0.0	0.5142	0.3643	0.6128	0.5035	0.8523	0.7453	0.7248	0.5885	0.0	0.0	0.4679	0.3429	0.3457	0.2623	0.5808	0.4045	0.2870	0.2606	0.6795	0.5014	0.8525	0.7381	0.9723	0.9355	0.7143	0.6309	0.0	0.0	0.6018	0.4166	0.0	0.0	0.7005	0.5535	0.3759	0.2465	0.0	0.0	0.7995	0.6740	0.6215	0.4400	0.9029	0.8031	0.0	0.0	0.8257	0.6465	0.5561	0.4781	0.8578	0.7371	0.4609	0.3572	0.5548	0.4644
1.5247	8.5227	750	1.7518	0.4350	0.5261	0.8188	0.6663	0.4424	0.0	0.0	0.7161	0.5921	0.6493	0.4981	0.8907	0.7577	0.8622	0.7597	0.7655	0.6298	0.7968	0.6739	0.0	0.0	0.6074	0.4314	0.0007	0.0007	0.0377	0.0376	0.9025	0.7946	0.0	0.0	0.3015	0.2309	0.9386	0.8433	0.9140	0.7810	0.9278	0.8216	0.0228	0.0192	0.7366	0.6193	0.8486	0.7361	0.2523	0.2274	0.6016	0.4953	0.0	0.0	0.5214	0.3686	0.5761	0.4885	0.8635	0.7524	0.7393	0.5908	0.0	0.0	0.5127	0.3698	0.3750	0.2777	0.5604	0.4056	0.2875	0.2671	0.6182	0.4901	0.8603	0.7434	0.9759	0.9390	0.7488	0.6711	0.0	0.0	0.5286	0.3854	0.0	0.0	0.7347	0.5366	0.4034	0.2692	0.0	0.0	0.8108	0.6822	0.5975	0.4477	0.9002	0.8097	0.0	0.0	0.7853	0.6437	0.5719	0.4936	0.8754	0.7391	0.4672	0.3655	0.6027	0.4934
1.5030	10.2273	900	1.7591	0.4375	0.5292	0.8200	0.6528	0.3962	0.0	0.0	0.7359	0.5948	0.6475	0.4863	0.8811	0.7654	0.8714	0.7623	0.7702	0.6310	0.7778	0.6554	0.0020	0.0020	0.6547	0.4233	0.0	0.0	0.0526	0.0519	0.9016	0.7990	0.0	0.0	0.3784	0.2541	0.9375	0.8448	0.9098	0.7870	0.9318	0.8338	0.0364	0.0288	0.7312	0.6206	0.8469	0.7405	0.3046	0.2287	0.6042	0.5055	0.0625	0.0590	0.4752	0.3603	0.5728	0.4990	0.8800	0.7517	0.7486	0.5993	0.0	0.0	0.4594	0.3574	0.3670	0.2867	0.5731	0.4067	0.3112	0.2800	0.6364	0.4865	0.8628	0.7481	0.9754	0.9379	0.8116	0.6813	0.0	0.0	0.5794	0.4195	0.0	0.0	0.6859	0.5465	0.3597	0.2640	0.0	0.0	0.7919	0.6786	0.5173	0.4171	0.8964	0.8115	0.0043	0.0043	0.8032	0.6615	0.5594	0.4772	0.8592	0.7401	0.5082	0.3815	0.5871	0.4842
1.4878	11.9318	1050	1.7587	0.4443	0.5333	0.8220	0.6410	0.4534	0.0	0.0	0.7198	0.6076	0.6592	0.4968	0.8767	0.7619	0.8673	0.7656	0.7730	0.6306	0.7623	0.6756	0.0078	0.0074	0.6262	0.4327	0.0552	0.0547	0.0821	0.0796	0.9052	0.8006	0.0	0.0	0.3560	0.2650	0.9393	0.8452	0.9111	0.7868	0.9279	0.8294	0.0337	0.0273	0.7243	0.6211	0.8559	0.7433	0.3035	0.2180	0.6411	0.5189	0.0676	0.0646	0.4799	0.3627	0.6125	0.5151	0.8833	0.7531	0.7479	0.6037	0.0	0.0	0.4667	0.3620	0.3773	0.2851	0.5570	0.4073	0.2921	0.2681	0.6186	0.4826	0.8532	0.7510	0.9742	0.9389	0.7981	0.6705	0.0	0.0	0.5839	0.4256	0.0	0.0	0.7143	0.5639	0.3564	0.2592	0.0	0.0	0.8101	0.6850	0.5650	0.4234	0.8946	0.8151	0.0436	0.0435	0.8234	0.6931	0.5984	0.4943	0.8596	0.7419	0.4865	0.3786	0.5961	0.4913
1.4774	13.6364	1200	1.7623	0.4532	0.5406	0.8224	0.6755	0.4518	0.0	0.0	0.7190	0.6032	0.6659	0.5084	0.8853	0.7646	0.8605	0.7578	0.7609	0.6274	0.7714	0.6618	0.0209	0.0189	0.6176	0.4249	0.0777	0.0771	0.0887	0.0831	0.9012	0.8021	0.0	0.0	0.3660	0.3238	0.9401	0.8457	0.9134	0.7870	0.9191	0.8267	0.0206	0.0165	0.7480	0.6279	0.8541	0.7449	0.2601	0.2271	0.6014	0.5052	0.1274	0.1127	0.4793	0.3624	0.5963	0.5059	0.8822	0.7548	0.7413	0.6046	0.0	0.0	0.4709	0.3654	0.3889	0.2853	0.5811	0.4103	0.2984	0.2706	0.6261	0.4917	0.8534	0.7521	0.9745	0.9388	0.7361	0.6369	0.0	0.0	0.5831	0.4231	0.0	0.0	0.6938	0.5581	0.3082	0.2441	0.0	0.0	0.8167	0.6844	0.5244	0.4070	0.9011	0.8107	0.4492	0.4449	0.8354	0.7040	0.5684	0.4837	0.8578	0.7432	0.5182	0.3877	0.6330	0.4985

Framework versions

• Transformers 5.0.0
• Pytorch 2.9.1+cu128
• Datasets 4.5.0
• Tokenizers 0.22.2