Segformer-b2-human-parse-24 Open-source Model - Accurately Implement Precise Segmentation of Human Body Parts and Clothing

Segformer B2 Human Parse 24

Developed by yolo12138

This model is a clothing segmentation model based on the SegFormer architecture, specifically designed for fine segmentation of human body parts and clothing.

Image Segmentation

Transformers

Open Source License:MIT #Clothing Part Segmentation #High-Precision Human Parsing #Multi-Label Semantic Segmentation

Downloads 27

Release Time : 12/8/2023

Model Overview

This is an image segmentation model based on the SegFormer-b2 architecture, specifically designed for fine segmentation of human body parts and clothing. The model was fine-tuned on the human_parsing_29_mix dataset and can recognize 24 different human body parts and clothing categories.

Model Features

Fine Human Part Segmentation

Can recognize 24 different human body parts and clothing categories, including detailed areas such as hair, face, arms, and legs.

High-Precision Segmentation

Achieves a mean Intersection over Union (IoU) of 0.6023 and an overall accuracy of 0.9780 on the evaluation dataset.

Clothing Recognition

Can distinguish between different types of clothing, such as tops, dresses, and outerwear.

Model Capabilities

Human Part Segmentation

Clothing Recognition

Image Semantic Segmentation

Use Cases

Fashion and Retail

Virtual Try-On

Used for virtual try-on features on e-commerce platforms, accurately identifying user body parts and existing clothing.

Clothing Recommendation

Provides personalized clothing recommendations based on analysis of user attire.

Human-Computer Interaction

Augmented Reality Applications

Accurately identifies user body parts in AR applications to enable more natural interactions.

🚀 segformer-b2-human-parse-24

This model is a fine - tuned version of mattmdjaga/segformer_b2_clothes on the human_parsing_29_mix dataset, achieving high - performance results in image segmentation.

🚀 Quick Start

This model is a fine - tuned version of mattmdjaga/segformer_b2_clothes on the human_parsing_29_mix dataset. It achieves the following results on the evaluation set:

Loss: 0.0818
Mean Iou: 0.6023
Mean Accuracy: 0.6321
Overall Accuracy: 0.9780
Accuracy Background: 0.9969
Accuracy Hat: nan
Accuracy Hair: 0.9646
Accuracy Glove: 0.0
Accuracy Glasses: 0.0
Accuracy Upper Only Torso Region: 0.9747
Accuracy Dresses Only Torso Region: 0.4939
Accuracy Coat Only Torso Region: 0.0039
Accuracy Socks: 0.0
Accuracy Left Pants: 0.9604
Accuracy Right Patns: 0.9646
Accuracy Skin Around Neck Region: 0.9585
Accuracy Scarf: nan
Accuracy Skirts: 0.8904
Accuracy Face: 0.9796
Accuracy Left Arm: 0.9703
Accuracy Right Arm: 0.9700
Accuracy Left Leg: 0.9267
Accuracy Right Leg: 0.9297
Accuracy Left Shoe: 0.0
Accuracy Right Shoe: 0.0
Accuracy Left Sleeve For Upper: 0.9462
Accuracy Right Sleeve For Upper: 0.9517
Accuracy Bag: 0.0234
Iou Background: 0.9941
Iou Hat: nan
Iou Hair: 0.9268
Iou Glove: 0.0
Iou Glasses: 0.0
Iou Upper Only Torso Region: 0.9351
Iou Dresses Only Torso Region: 0.4059
Iou Coat Only Torso Region: 0.0035
Iou Socks: 0.0
Iou Left Pants: 0.9232
Iou Right Patns: 0.9217
Iou Skin Around Neck Region: 0.9227
Iou Scarf: nan
Iou Skirts: 0.7887
Iou Face: 0.9582
Iou Left Arm: 0.9436
Iou Right Arm: 0.9426
Iou Left Leg: 0.8836
Iou Right Leg: 0.8767
Iou Left Shoe: 0.0
Iou Right Shoe: 0.0
Iou Left Sleeve For Upper: 0.9005
Iou Right Sleeve For Upper: 0.9012
Iou Bag: 0.0232

📚 Documentation

Model description

"id2label": {
    "0": "background",
    "1": "hat",
    "2": "hair",
    "3": "glove",
    "4": "glasses",
    "5": "upper_only_torso_region",
    "6": "dresses_only_torso_region",
    "7": "coat_only_torso_region",
    "8": "socks",
    "9": "left_pants",
    "10": "right_patns",
    "11": "skin_around_neck_region",
    "12": "scarf",
    "13": "skirts",
    "14": "face",
    "15": "left_arm",
    "16": "right_arm",
    "17": "left_leg",
    "18": "right_leg",
    "19": "left_shoe",
    "20": "right_shoe",
    "21": "left_sleeve_for_upper",
    "22": "right_sleeve_for_upper",
    "23": "bag"
}

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: 6e - 05
train_batch_size: 16
eval_batch_size: 12
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e - 08
lr_scheduler_type: linear
num_epochs: 8

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Hat	Accuracy Hair	Accuracy Upper Only Torso Region	Accuracy Dresses Only Torso Region	Accuracy Coat Only Torso Region	Accuracy Left Pants	Accuracy Right Patns	Accuracy Skin Around Neck Region	Accuracy Scarf	Accuracy Skirts	Accuracy Face	Accuracy Left Arm	Accuracy Right Arm	Accuracy Left Leg	Accuracy Right Leg	Accuracy Left Sleeve For Upper	Accuracy Right Sleeve For Upper	Accuracy Bag	Iou Background	Iou Hat	Iou Hair	Iou Upper Only Torso Region	Iou Dresses Only Torso Region	Iou Coat Only Torso Region	Iou Left Pants	Iou Right Patns	Iou Skin Around Neck Region	Iou Scarf	Iou Skirts	Iou Face	Iou Left Arm	Iou Right Arm	Iou Left Leg	Iou Right Leg	Iou Left Sleeve For Upper	Iou Right Sleeve For Upper	Iou Bag
0.0652	1.62	1000	0.0802	0.5857	0.6166	0.9737	0.9963	nan	0.9490	0.9801	0.4034	0.0	0.9487	0.9574	0.9272	nan	0.8783	0.9782	0.9628	0.9534	0.8874	0.9012	0.9227	0.9197	0.0	0.9926	nan	0.9117	0.9217	0.3541	0.0	0.9084	0.9073	0.8963	nan	0.7766	0.9455	0.9210	0.9191	0.8405	0.8496	0.8673	0.8728	0.0
0.061	3.23	2000	0.0843	0.5977	0.6335	0.9747	0.9967	nan	0.9580	0.9657	0.5733	0.1504	0.9591	0.9600	0.9497	nan	0.8169	0.9789	0.9667	0.9645	0.8906	0.9165	0.9444	0.9445	0.0003	0.9935	nan	0.9199	0.9273	0.4058	0.1206	0.9131	0.9082	0.9128	nan	0.7330	0.9527	0.9355	0.9343	0.8534	0.8651	0.8860	0.8879	0.0003
0.0653	4.85	3000	0.0823	0.6000	0.6295	0.9775	0.9967	nan	0.9621	0.9780	0.4991	0.0044	0.9587	0.9649	0.9562	nan	0.8842	0.9769	0.9692	0.9651	0.9198	0.9273	0.9422	0.9415	0.0037	0.9939	nan	0.9247	0.9341	0.4136	0.0042	0.9202	0.9193	0.9193	nan	0.7899	0.9563	0.9403	0.9388	0.8745	0.8741	0.8963	0.8970	0.0037
0.0402	6.46	4000	0.0818	0.6023	0.6321	0.9780	0.9969	nan	0.9646	0.9747	0.4939	0.0039	0.9604	0.9646	0.9585	nan	0.8904	0.9796	0.9703	0.9700	0.9267	0.9297	0.9462	0.9517	0.0234	0.9941	nan	0.9268	0.9351	0.4059	0.0035	0.9232	0.9217	0.9227	nan	0.7887	0.9582	0.9436	0.9426	0.8836	0.8767	0.9005	0.9012	0.0232

Framework versions

Transformers 4.35.2
Pytorch 2.1.1
Datasets 2.15.0
Tokenizers 0.15.0

📄 License

This project is licensed under the MIT 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