Segformer - B5 Finetuned Human Parsing Open - Source Vision Model: Free Deployment for Precise Analysis of Human Body Parts and Clothing

Segformer B5 Finetuned Human Parsing

Developed by matei-dorian

A vision segmentation model based on MIT-B5 architecture, specialized in pixel-level parsing of human body parts and clothing

Image Segmentation

Transformers

Open Source License:Other #Human body part segmentation #High-precision apparel recognition #Multi-category semantic segmentation

Downloads 4,066

Release Time : 5/2/2023

Model Overview

This model is a fine-tuned version of NVIDIA's MIT-B5 architecture for human parsing tasks, capable of accurately segmenting human body parts (e.g., face, limbs) and clothing accessories (e.g., hats, shoes, bags) in images

Model Features

High-precision part segmentation

Achieves over 90% accuracy in key areas such as face and hair

Multi-category recognition

Supports simultaneous recognition of 20+ human body parts and clothing accessories

Efficient fine-tuning architecture

Based on the pre-trained SegFormer-B5 model, excellent performance can be achieved with minimal fine-tuning data

Model Capabilities

Human body part segmentation

Clothing accessory recognition

Pixel-level image parsing

Multi-category semantic segmentation

Use Cases

Fashion analysis

Virtual fitting effect analysis

Identifies various components of the user's clothing

Top recognition accuracy 85.53%, pants 89.13%

Human-computer interaction

Augmented reality applications

Precisely locates human body parts for virtual overlays

Face recognition accuracy 90.94%, arm recognition 85%+

🚀 segformer-b5-finetuned-human-parsing

This model is a fine - tuned version of nvidia/mit-b5 on the None dataset. It specializes in image segmentation, particularly for human parsing, and offers high - precision results in relevant tasks.

🚀 Quick Start

This model is ready to use for human parsing tasks. You can directly load it from the Hugging Face model hub and start making inferences.

📚 Documentation

Model Performance

It achieves the following results on the evaluation set:

Loss: 0.2292
Mean Iou: 0.6258
Mean Accuracy: 0.7547
Overall Accuracy: 0.8256
Accuracy Background: nan
Accuracy Hat: 0.8561
Accuracy Hair: 0.8974
Accuracy Sunglasses: 0.7540
Accuracy Upper - clothes: 0.8553
Accuracy Skirt: 0.7026
Accuracy Pants: 0.8913
Accuracy Dress: 0.7525
Accuracy Belt: 0.4251
Accuracy Left - shoe: 0.6014
Accuracy Right - shoe: 0.6374
Accuracy Face: 0.9094
Accuracy Left - leg: 0.8452
Accuracy Right - leg: 0.8343
Accuracy Left - arm: 0.8506
Accuracy Right - arm: 0.8287
Accuracy Bag: 0.8232
Accuracy Scarf: 0.3662
Iou Background: 0.0
Iou Hat: 0.7625
Iou Hair: 0.8171
Iou Sunglasses: 0.6400
Iou Upper - clothes: 0.7700
Iou Skirt: 0.6211
Iou Pants: 0.7788
Iou Dress: 0.5512
Iou Belt: 0.3564
Iou Left - shoe: 0.5032
Iou Right - shoe: 0.5381
Iou Face: 0.8294
Iou Left - leg: 0.7412
Iou Right - leg: 0.7591
Iou Left - arm: 0.7579
Iou Right - arm: 0.7705
Iou Bag: 0.7729
Iou Scarf: 0.2956

Training Procedure

Training Hyperparameters

The following hyperparameters were used during training:

learning_rate: 6e - 05
train_batch_size: 2
eval_batch_size: 2
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
num_epochs: 15

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 Left - shoe	Accuracy Right - shoe	Accuracy Face	Accuracy Left - leg	Accuracy Right - leg	Accuracy Left - arm	Accuracy Right - arm	Accuracy Bag	Iou Hat	Iou Hair	Iou Sunglasses	Iou Upper - clothes	Iou Skirt	Iou Pants	Iou Dress	Iou Left - shoe	Iou Right - shoe	Iou Face	Iou Left - leg	Iou Right - leg	Iou Left - arm	Iou Right - arm	Iou Bag
1.1597	0.04	20	1.5815	0.1179	0.1991	0.4296	nan	0.0060	0.6905	0.0	0.7657	0.0108	0.6431	0.2946	0.0288	0.0366	0.1480	0.0025	0.5692	0.0096	0.0259	0.1537	0.0051	0.4253	0.0	0.5199	0.0103	0.3388	0.1700	0.0258	0.0338	0.0895	0.0025	0.3162	0.0094	0.0253	0.1495
0.6963	0.08	40	0.8073	0.1759	0.2719	0.4628	nan	0.0015	0.8699	0.0	0.4736	0.4932	0.5141	0.6775	0.0062	0.1038	0.5301	0.0916	0.5071	0.0092	0.0549	0.2889	0.0015	0.6169	0.0	0.4242	0.2202	0.3522	0.2251	0.0062	0.0904	0.4914	0.0852	0.3160	0.0092	0.0541	0.2731
0.5786	0.12	60	0.6136	0.2538	0.3642	0.4679	nan	0.0180	0.8122	0.0	0.1998	0.0000	0.6621	0.8592	0.1440	0.2772	0.8381	0.4032	0.6068	0.4182	0.3097	0.6434	0.0179	0.6760	0.0	0.1951	0.0000	0.5471	0.2218	0.1147	0.2032	0.6403	0.3189	0.4204	0.3505	0.2947	0.5676
0.324	0.16	80	0.4282	0.2893	0.4044	0.6041	nan	0.0147	0.7890	0.0	0.8222	0.7984	0.6646	0.1038	0.0896	0.3308	0.8277	0.4099	0.6839	0.2401	0.5474	0.5521	0.0147	0.6800	0.0	0.6159	0.3049	0.5913	0.0938	0.0802	0.2394	0.6598	0.3178	0.4504	0.2288	0.4189	0.5113
0.297	0.2	100	0.4020	0.3034	0.4230	0.6332	nan	0.0048	0.8076	0.0080	0.9042	0.6567	0.8036	0.0317	0.0481	0.5298	0.7728	0.2589	0.7232	0.5941	0.3839	0.6643	0.0048	0.6708	0.0080	0.6300	0.3836	0.5929	0.0314	0.0441	0.3152	0.6726	0.2420	0.4745	0.4532	0.3631	0.5759
0.2608	0.24	120	0.3538	0.3444	0.4554	0.6504	nan	0.2922	0.8078	0.0753	0.8472	0.0425	0.6961	0.6197	0.2550	0.3074	0.8020	0.5636	0.6895	0.3779	0.6930	0.6734	0.2757	0.6940	0.0747	0.6457	0.0419	0.6098	0.3611	0.1849	0.2412	0.7038	0.4513	0.5038	0.3439	0.4760	0.5915
0.3306	0.28	140	0.3281	0.3562	0.4736	0.6560	nan	0.4111	0.8576	0.1953	0.8081	0.6916	0.7888	0.3489	0.0809	0.3612	0.8132	0.0622	0.7078	0.6328	0.5437	0.7482	0.3895	0.7227	0.1857	0.6777	0.3750	0.6015	0.2749	0.0740	0.2602	0.7070	0.0612	0.4348	0.5114	0.4966	0.6385
0.364	0.32	160	0.3368	0.3689	0.4836	0.6531	nan	0.3898	0.8453	0.1743	0.9269	0.2493	0.7922	0.0842	0.4874	0.2384	0.8116	0.6226	0.5731	0.6049	0.6620	0.7597	0.3746	0.7246	0.1690	0.6015	0.1998	0.5942	0.0786	0.2682	0.1904	0.7015	0.4781	0.4781	0.5452	0.5804	0.6562
0.635	0.36	180	0.3092	0.3699	0.4903	0.6319	nan	0.4996	0.8387	0.2136	0.6184	0.0129	0.7920	0.8199	0.1895	0.3028	0.8307	0.7258	0.3386	0.7480	0.6543	0.7511	0.4613	0.7126	0.2042	0.5589	0.0128	0.6658	0.3529	0.1622	0.2426	0.7363	0.4646	0.3144	0.5794	0.5575	0.6321
0.1464	0.4	200	0.3306	0.3809	0.5041	0.6544	nan	0.6110	0.8337	0.2420	0.8913	0.8862	0.6492	0.0004	0.2888	0.2949	0.8514	0.4630	0.7751	0.7020	0.5429	0.5386	0.5329	0.7348	0.2331	0.6567	0.3661	0.5769	0.0004	0.2221	0.2333	0.7431	0.4133	0.5478	0.5718	0.5125	0.5107
0.2257	0.44	220	0.2751	0.4089	0.5400	0.6752	nan	0.6851	0.8458	0.4204	0.7241	0.1085	0.7997	0.7657	0.2458	0.4039	0.8858	0.7863	0.3199	0.7405	0.6974	0.7508	0.5815	0.7437	0.3776	0.6458	0.1033	0.6526	0.3966	0.2027	0.3078	0.7438	0.4680	0.2966	0.6204	0.5942	0.6260
0.3069	0.48	240	0.2614	0.4163	0.5499	0.6868	nan	0.6246	0.8571	0.3130	0.7765	0.8266	0.7786	0.3212	0.3560	0.3736	0.8579	0.1780	0.8761	0.7423	0.7693	0.6970	0.5597	0.7370	0.2931	0.6733	0.4032	0.6889	0.2487	0.2662	0.2901	0.7425	0.1724	0.4957	0.6373	0.6376	0.6470
0.1454	0.52	260	0.2563	0.4316	0.5610	0.6965	nan	0.6707	0.8388	0.5572	0.7616	0.3854	0.7280	0.7114	0.1934	0.3621	0.8718	0.7860	0.6140	0.7403	0.5340	0.7820	0.5710	0.7446	0.4497	0.6637	0.3125	0.6624	0.4219	0.1731	0.2862	0.7295	0.5339	0.5054	0.5742	0.4967	0.6449
0.1522	0.56	280	0.2521	0.4327	0.5567	0.7138	nan	0.5098	0.9135	0.3399	0.8898	0.5537	0.7508	0.2922	0.3367	0.2484	0.8388	0.7460	0.7191	0.7496	0.7996	0.7753	0.4902	0.7541	0.3196	0.6924	0.3853	0.6261	0.2512	0.2575	0.2171	0.7393	0.5563	0.5633	0.6403	0.6335	0.6621
0.1872	0.6	300	0.2472	0.4432	0.5674	0.7252	nan	0.6434	0.8437	0.4432	0.8047	0.3844	0.7534	0.7336	0.2764	0.2778	0.8632	0.7863	0.6434	0.7496	0.7996	0.7753	0.5710	0.7446	0.4497	0.6637	0.3125	0.6624	0.4219	0.1731	0.2862	0.7295	0.5339	0.5054	0.5742	0.4967	0.6449

📄 License

This model is released under the other license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご