finetune-instance-segmentation-ade20k-mini-mask2former_3

This model is a fine-tuned version of facebook/mask2former-swin-tiny-coco-instance on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 21.8296
• Map: 0.6941
• Map 50: 0.8672
• Map 75: 0.7324
• Map Small: 0.1837
• Map Medium: 0.533
• Map Large: 0.8476
• Mar 1: 0.5536
• Mar 10: 0.7509
• Mar 100: 0.7971
• Mar Small: 0.4162
• Mar Medium: 0.7363
• Mar Large: 0.9159
• Map Background: 0.9263
• Mar 100 Background: 0.9435
• Map Building: 0.462
• Mar 100 Building: 0.6506

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: 1e-05
• train_batch_size: 64
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 30.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Map	Map 50	Map 75	Map Small	Map Medium	Map Large	Mar 1	Mar 10	Mar 100	Mar Small	Mar Medium	Mar Large	Map Background	Mar 100 Background	Map Building	Mar 100 Building
113.7971	1.0	18	54.5754	0.0104	0.0275	0.0056	0.0007	0.0092	0.0318	0.0158	0.1576	0.3476	0.0456	0.1433	0.3825	0.0169	0.5786	0.004	0.1165
46.1207	2.0	36	41.7255	0.1211	0.2383	0.1079	0.0297	0.1815	0.3339	0.1421	0.4984	0.6046	0.2304	0.5813	0.7349	0.088	0.7217	0.1542	0.4874
37.5319	3.0	54	35.0953	0.3438	0.5097	0.3674	0.0693	0.3147	0.5321	0.4204	0.6597	0.7112	0.2853	0.62	0.8425	0.4114	0.8886	0.2762	0.5339
32.2408	4.0	72	31.1007	0.4725	0.6521	0.5066	0.0955	0.3797	0.6318	0.5057	0.6913	0.7412	0.3187	0.6466	0.8723	0.6119	0.9191	0.3331	0.5634
29.0339	5.0	90	28.8429	0.5993	0.7915	0.6487	0.1077	0.4116	0.7603	0.5213	0.7062	0.7536	0.3345	0.6634	0.8849	0.8423	0.9273	0.3563	0.5798
27.0289	6.0	108	27.2669	0.6292	0.8239	0.6783	0.1232	0.4349	0.7877	0.5305	0.7117	0.7585	0.3505	0.6724	0.889	0.883	0.9253	0.3753	0.5916
25.7301	7.0	126	26.2551	0.6392	0.8322	0.6873	0.1325	0.4502	0.7974	0.532	0.7143	0.7647	0.3629	0.69	0.8908	0.8905	0.9248	0.3879	0.6047
24.7686	8.0	144	25.5236	0.6473	0.8391	0.6943	0.1394	0.4632	0.8027	0.5368	0.718	0.7679	0.3667	0.6967	0.8943	0.8956	0.9257	0.399	0.6102
24.1245	9.0	162	24.9991	0.6529	0.8439	0.6976	0.1428	0.4749	0.8062	0.539	0.7193	0.7697	0.3736	0.7027	0.895	0.8974	0.9234	0.4084	0.6161
23.5334	10.0	180	24.7225	0.6582	0.8458	0.7004	0.1506	0.4812	0.8093	0.5416	0.724	0.7744	0.3799	0.7082	0.8968	0.9016	0.9282	0.4149	0.6206
23.0677	11.0	198	24.3978	0.6632	0.849	0.7049	0.1521	0.4852	0.8161	0.5432	0.7258	0.7761	0.3796	0.7115	0.9006	0.9071	0.9289	0.4193	0.6233
22.597	12.0	216	24.0400	0.6675	0.8523	0.707	0.157	0.4926	0.8196	0.5438	0.7295	0.7784	0.3842	0.7136	0.9032	0.9094	0.9301	0.4255	0.6267
22.2306	13.0	234	23.8523	0.6705	0.8544	0.7078	0.1584	0.4932	0.822	0.5466	0.7306	0.7817	0.3844	0.7202	0.9055	0.9134	0.9335	0.4276	0.6299
21.8893	14.0	252	23.6264	0.6735	0.8559	0.7123	0.1607	0.5	0.8253	0.548	0.7344	0.7829	0.3922	0.7166	0.9055	0.9144	0.9349	0.4327	0.6308
21.6071	15.0	270	23.4112	0.676	0.8575	0.7141	0.1625	0.5027	0.8278	0.5492	0.7355	0.7836	0.3919	0.7178	0.9063	0.9167	0.9358	0.4353	0.6314
21.2754	16.0	288	23.1654	0.6782	0.8573	0.7177	0.1649	0.5109	0.8295	0.5478	0.7382	0.7869	0.3969	0.7282	0.9077	0.9166	0.9353	0.4398	0.6385
21.0393	17.0	306	23.0104	0.6796	0.8592	0.7186	0.1682	0.5086	0.8329	0.5484	0.7395	0.7868	0.396	0.7247	0.9095	0.9184	0.9365	0.4408	0.6371
20.8053	18.0	324	22.8456	0.683	0.8598	0.7218	0.1705	0.5171	0.8342	0.5503	0.7417	0.7895	0.4014	0.7308	0.9109	0.9195	0.9369	0.4465	0.6422
20.5247	19.0	342	22.7148	0.6829	0.8614	0.7207	0.1696	0.5175	0.8339	0.5499	0.7417	0.7895	0.4034	0.729	0.9102	0.9201	0.9373	0.4457	0.6417
20.4118	20.0	360	22.5014	0.6844	0.8605	0.7211	0.1705	0.5185	0.8375	0.5516	0.7432	0.7909	0.4025	0.7317	0.9109	0.9212	0.9394	0.4476	0.6425
20.1455	21.0	378	22.4692	0.6863	0.864	0.7234	0.1715	0.5212	0.8375	0.5513	0.7445	0.7924	0.4055	0.7332	0.9116	0.9226	0.9406	0.4501	0.6443
19.904	22.0	396	22.4235	0.6877	0.8654	0.7244	0.1737	0.5248	0.8411	0.5526	0.7432	0.792	0.4058	0.7314	0.9133	0.9227	0.9396	0.4526	0.6445
19.6888	23.0	414	22.3299	0.6902	0.8652	0.7264	0.1769	0.5283	0.8421	0.5531	0.7446	0.7929	0.409	0.7326	0.9143	0.924	0.9392	0.4563	0.6466
19.4733	24.0	432	22.2339	0.69	0.8649	0.7267	0.1763	0.5277	0.8426	0.5527	0.7463	0.7937	0.4107	0.7318	0.9142	0.9243	0.9408	0.4558	0.6465
19.3223	25.0	450	22.1182	0.6902	0.8656	0.7275	0.1793	0.5286	0.843	0.5525	0.7469	0.7939	0.4105	0.7342	0.9159	0.9234	0.9394	0.457	0.6484
19.1744	26.0	468	22.0561	0.6915	0.8678	0.7304	0.1798	0.5284	0.8448	0.5524	0.7457	0.7936	0.4125	0.7326	0.9133	0.9238	0.9398	0.4592	0.6474
19.002	27.0	486	22.0421	0.6932	0.8676	0.7294	0.1811	0.5334	0.8447	0.5522	0.7485	0.7966	0.4156	0.7399	0.9151	0.9251	0.941	0.4612	0.6522
18.8741	28.0	504	21.9818	0.6939	0.8683	0.7334	0.1836	0.5328	0.847	0.553	0.7494	0.7968	0.4165	0.7361	0.9173	0.9263	0.9421	0.4615	0.6514
19.2285	28.3429	510	21.8296	0.6941	0.8672	0.7324	0.1837	0.533	0.8476	0.5536	0.7509	0.7971	0.4162	0.7363	0.9159	0.9263	0.9435	0.462	0.6506

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0