mit-b0-finetuned-human-parsing-dataset

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

• Loss: 0.1612
• Mean Iou: 0.5450
• Mean Accuracy: 0.6607
• Overall Accuracy: 0.8160
• Accuracy Background: nan
• Accuracy Hat: 0.5935
• Accuracy Hair: 0.8675
• Accuracy Sunglasses: 0.1278
• Accuracy Upper-clothes: 0.8806
• Accuracy Skirt: 0.7150
• Accuracy Pants: 0.8529
• Accuracy Dress: 0.8186
• Accuracy Belt: 0.0817
• Accuracy Left-shoe: 0.6562
• Accuracy Right-shoe: 0.6193
• Accuracy Face: 0.8987
• Accuracy Left-leg: 0.8838
• Accuracy Right-leg: 0.8541
• Accuracy Left-arm: 0.8193
• Accuracy Right-arm: 0.8202
• Accuracy Bag: 0.7409
• Accuracy Scarf: 0.0012
• Iou Background: 0.0
• Iou Hat: 0.5417
• Iou Hair: 0.7745
• Iou Sunglasses: 0.1273
• Iou Upper-clothes: 0.7733
• Iou Skirt: 0.6469
• Iou Pants: 0.7596
• Iou Dress: 0.6192
• Iou Belt: 0.0773
• Iou Left-shoe: 0.5307
• Iou Right-shoe: 0.5156
• Iou Face: 0.8002
• Iou Left-leg: 0.7577
• Iou Right-leg: 0.7632
• Iou Left-arm: 0.7325
• Iou Right-arm: 0.7315
• Iou Bag: 0.6578
• Iou Scarf: 0.0012

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: 7e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Hat	Accuracy Hair	Accuracy Sunglasses	Accuracy Upper-clothes	Accuracy Skirt	Accuracy Pants	Accuracy Dress	Accuracy Belt	Accuracy Left-shoe	Accuracy Right-shoe	Accuracy Face	Accuracy Left-leg	Accuracy Right-leg	Accuracy Left-arm	Accuracy Right-arm	Accuracy Bag	Accuracy Scarf	Iou Background	Iou Hat	Iou Hair	Iou Sunglasses	Iou Upper-clothes	Iou Skirt	Iou Pants	Iou Dress	Iou Belt	Iou Left-shoe	Iou Right-shoe	Iou Face	Iou Left-leg	Iou Right-leg	Iou Left-arm	Iou Right-arm	Iou Bag	Iou Scarf
0.1738	1.0	200	0.2036	0.4669	0.5844	0.7641	nan	0.2353	0.8366	0.0	0.8116	0.6698	0.7972	0.8264	0.0	0.5317	0.4734	0.8657	0.8419	0.7916	0.7738	0.7692	0.7112	0.0	0.0	0.2283	0.7279	0.0	0.7127	0.5858	0.7043	0.5625	0.0	0.4160	0.3786	0.7652	0.6935	0.6938	0.6787	0.6674	0.5891	0.0
0.184	2.0	400	0.1841	0.4970	0.6199	0.7940	nan	0.4453	0.8607	0.0	0.8745	0.7569	0.8160	0.7201	0.0	0.5756	0.5385	0.9054	0.8440	0.8553	0.8249	0.8242	0.6971	0.0	0.0	0.4153	0.7599	0.0	0.7467	0.6348	0.7162	0.5777	0.0	0.4517	0.4229	0.7710	0.7025	0.7143	0.7096	0.7060	0.6183	0.0
0.1793	3.0	600	0.1717	0.5121	0.6276	0.8018	nan	0.5648	0.8591	0.0	0.8920	0.7414	0.8757	0.7207	0.0	0.6178	0.5797	0.8696	0.8117	0.8442	0.7867	0.7884	0.7170	0.0	0.0	0.4805	0.7622	0.0	0.7497	0.6576	0.7450	0.5970	0.0	0.4793	0.4538	0.7821	0.7290	0.7402	0.7075	0.7070	0.6269	0.0
0.3023	4.0	800	0.1753	0.5129	0.6313	0.7953	nan	0.5461	0.8778	0.0	0.8113	0.7911	0.8080	0.8468	0.0	0.6061	0.5468	0.8959	0.8538	0.8359	0.8053	0.8009	0.7055	0.0	0.0	0.4921	0.7589	0.0	0.7408	0.6533	0.7325	0.5989	0.0	0.4843	0.4550	0.7872	0.7399	0.7417	0.7114	0.7089	0.6265	0.0
0.1041	5.0	1000	0.1655	0.5235	0.6388	0.8078	nan	0.6147	0.8667	0.0025	0.8768	0.7477	0.8536	0.7777	0.0022	0.5801	0.5814	0.8896	0.8580	0.8658	0.8238	0.8236	0.6964	0.0	0.0	0.5389	0.7662	0.0025	0.7582	0.6485	0.7581	0.6070	0.0022	0.4840	0.4767	0.7900	0.7534	0.7572	0.7267	0.7204	0.6336	0.0
0.1179	6.0	1200	0.1628	0.5312	0.6475	0.8111	nan	0.5886	0.8725	0.0326	0.8560	0.7353	0.8538	0.8384	0.0221	0.6322	0.5871	0.9038	0.8580	0.8579	0.8263	0.8279	0.7142	0.0	0.0	0.5293	0.7663	0.0326	0.7629	0.6531	0.7624	0.6189	0.0217	0.5135	0.4931	0.7930	0.7599	0.7641	0.7293	0.7224	0.6386	0.0
0.1323	7.0	1400	0.1619	0.5390	0.6531	0.8129	nan	0.6147	0.8846	0.0754	0.8677	0.7143	0.8672	0.8331	0.0484	0.6528	0.6319	0.8896	0.8392	0.8467	0.8096	0.8072	0.7201	0.0	0.0	0.5489	0.7726	0.0753	0.7693	0.6392	0.7660	0.6163	0.0468	0.5264	0.5172	0.7976	0.7612	0.7666	0.7314	0.7245	0.6434	0.0
0.1235	8.0	1600	0.1612	0.5450	0.6607	0.8160	nan	0.5935	0.8675	0.1278	0.8806	0.7150	0.8529	0.8186	0.0817	0.6562	0.6193	0.8987	0.8838	0.8541	0.8193	0.8202	0.7409	0.0012	0.0	0.5417	0.7745	0.1273	0.7733	0.6469	0.7596	0.6192	0.0773	0.5307	0.5156	0.8002	0.7577	0.7632	0.7325	0.7315	0.6578	0.0012
0.093	9.0	1800	0.1621	0.5489	0.6621	0.8165	nan	0.6190	0.8752	0.1554	0.8799	0.7153	0.8561	0.8333	0.0834	0.6491	0.6138	0.8950	0.8622	0.8528	0.8119	0.8053	0.7411	0.0068	0.0	0.5590	0.7745	0.1544	0.7730	0.6432	0.7665	0.6144	0.0788	0.5308	0.5146	0.8012	0.7651	0.7687	0.7353	0.7305	0.6629	0.0068
0.1171	10.0	2000	0.1631	0.5504	0.6659	0.8147	nan	0.6300	0.8701	0.1681	0.8613	0.7188	0.8489	0.8576	0.0908	0.6585	0.6139	0.8979	0.8608	0.8580	0.8264	0.8141	0.7360	0.0096	0.0	0.5640	0.7730	0.1669	0.7699	0.6460	0.7657	0.6125	0.0854	0.5349	0.5162	0.8019	0.7652	0.7697	0.7366	0.7317	0.6573	0.0096

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2