SegFormer-B2-Cloth-Parse-9 Open-Source Clothing Segmentation Model - Accurately Identify and Segment Various Parts of Clothing

Segformer B2 Cloth Parse 9

Developed by yolo12138

A garment segmentation model based on the SegFormer architecture, fine-tuned on the cloth_parsing_mix dataset for precise identification and segmentation of different clothing parts.

Image Segmentation

Transformers

Open Source License:MIT #Garment Part Segmentation #High-Precision Semantic Segmentation #Fashion Image Parsing

Downloads 156

Release Time : 12/15/2023

Model Overview

This model is specifically designed for garment image segmentation tasks, capable of identifying and segmenting various components of clothing such as sleeves, pant legs, collars, etc. Suitable for fashion analysis, virtual try-on, and similar scenarios.

Model Features

High-Precision Garment Segmentation

Achieves high accuracy on multiple garment parts, particularly excelling in major components like upper body, pant legs, and sleeves.

Multi-Part Recognition

Capable of identifying and segmenting up to 11 different garment parts, including detailed components like inner/outer collars and left/right sleeves.

Efficient Training

Uses relatively small batch sizes (12) and moderate learning rates (1e-05) for efficient training, achieving good results in just 5 epochs.

Model Capabilities

Garment Image Segmentation

Multi-Part Recognition

Pixel-Level Classification

Use Cases

Fashion Technology

Virtual Try-On Systems

Used for virtual try-on features in online shopping platforms, precisely segmenting garment parts to achieve realistic fitting effects.

Fashion Design Analysis

Assists designers in analyzing garment structure and component proportions to optimize design solutions.

E-Commerce

Automatic Product Image Tagging

Automatically adds part labels to garment product images on e-commerce platforms, improving search and recommendation accuracy.

🚀 segformer-b2-cloth-parse-9

This model is a fine - tuned version of mattmdjaga/segformer_b2_clothes on the cloth_parsing_mix dataset. It offers high - performance image segmentation capabilities for clothing, achieving excellent results on the evaluation set.

🚀 Quick Start

This model is ready for use in image segmentation tasks. You can load it using relevant libraries and start making predictions.

📚 Documentation

Model Information

Property	Details
Model Type	Fine - tuned version of mattmdjaga/segformer_b2_clothes
Training Data	cloth_parsing_mix
Pipeline Tag	image - segmentation

Evaluation Results

This model achieves the following results on the evaluation set:

Loss: 0.0433
Mean Iou: 0.8611
Mean Accuracy: 0.9107
Overall Accuracy: 0.9846
Accuracy Background: 0.9964
Accuracy Upper Torso: 0.9857
Accuracy Left Pants: 0.9654
Accuracy Right Patns: 0.9664
Accuracy Skirts: 0.9065
Accuracy Left Sleeve: 0.9591
Accuracy Right Sleeve: 0.9662
Accuracy Outer Collar: 0.6491
Accuracy Inner Collar: 0.8015
Iou Background: 0.9923
Iou Upper Torso: 0.9655
Iou Left Pants: 0.9017
Iou Right Patns: 0.9085
Iou Skirts: 0.8749
Iou Left Sleeve: 0.9223
Iou Right Sleeve: 0.9289
Iou Outer Collar: 0.5394
Iou Inner Collar: 0.7160

Training Procedure

Training Hyperparameters

The following hyperparameters were used during training:

learning_rate: 1e - 05
train_batch_size: 12
eval_batch_size: 12
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
num_epochs: 5

Training Results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Upper Torso	Accuracy Left Pants	Accuracy Right Patns	Accuracy Skirts	Accuracy Left Sleeve	Accuracy Right Sleeve	Accuracy Outer Collar	Accuracy Inner Collar	Iou Background	Iou Upper Torso	Iou Left Pants	Iou Right Patns	Iou Skirts	Iou Left Sleeve	Iou Right Sleeve	Iou Outer Collar	Iou Inner Collar
0.1054	0.11	500	0.1180	0.7305	0.7971	0.9670	0.9902	0.9720	0.9654	0.9756	0.8036	0.9226	0.9289	0.0716	0.5444	0.9830	0.9234	0.8752	0.8765	0.7370	0.8236	0.8232	0.0703	0.4628
0.1033	0.22	1000	0.0851	0.7862	0.8418	0.9746	0.9924	0.9829	0.9665	0.9653	0.8491	0.9145	0.9226	0.3219	0.6608	0.9866	0.9424	0.8858	0.8875	0.8105	0.8538	0.8614	0.2833	0.5642
0.0944	0.32	1500	0.0713	0.8077	0.8595	0.9773	0.9941	0.9833	0.9566	0.9625	0.8924	0.9094	0.9181	0.4414	0.6774	0.9880	0.9481	0.8937	0.8950	0.8437	0.8668	0.8751	0.3629	0.5958
0.0746	0.43	2000	0.0683	0.8190	0.8770	0.9783	0.9941	0.9796	0.9652	0.9722	0.8656	0.9480	0.9562	0.4882	0.7236	0.9888	0.9497	0.9070	0.9127	0.8306	0.8790	0.8870	0.3945	0.6218
0.0548	0.54	2500	0.0666	0.8187	0.8713	0.9787	0.9951	0.9831	0.9580	0.9606	0.8651	0.9215	0.9453	0.4839	0.7293	0.9893	0.9514	0.8939	0.9006	0.8245	0.8812	0.8964	0.4010	0.6298
0.0728	0.65	3000	0.0591	0.8271	0.8806	0.9804	0.9945	0.9839	0.9624	0.9659	0.8982	0.9399	0.9430	0.4884	0.7493	0.9900	0.9551	0.8940	0.8966	0.8583	0.8930	0.9011	0.4100	0.6458
0.0505	0.75	3500	0.0648	0.8218	0.8745	0.9797	0.9947	0.9847	0.9858	0.9905	0.8402	0.9500	0.9587	0.4480	0.7178	0.9900	0.9534	0.9022	0.9037	0.8223	0.8944	0.9017	0.3881	0.6402
0.0601	0.86	4000	0.0568	0.8415	0.8951	0.9817	0.9952	0.9817	0.9632	0.9640	0.9170	0.9521	0.9541	0.5781	0.7508	0.9903	0.9576	0.9138	0.9199	0.8716	0.9010	0.9106	0.4562	0.6529
0.0438	0.97	4500	0.0569	0.8431	0.8925	0.9815	0.9947	0.9844	0.9764	0.9838	0.8870	0.9492	0.9595	0.5561	0.7416	0.9903	0.9560	0.9287	0.9370	0.8585	0.9000	0.9089	0.4524	0.6559
0.0617	1.08	5000	0.0529	0.8417	0.8933	0.9816	0.9952	0.9841	0.9602	0.9631	0.8922	0.9475	0.9533	0.5797	0.7642	0.9907	0.9571	0.9097	0.9126	0.8488	0.9044	0.9158	0.4687	0.6678
0.0452	1.19	5500	0.0557	0.8351	0.8935	0.9812	0.9949	0.9842	0.9644	0.9667	0.8781	0.9494	0.9604	0.5961	0.7471	0.9906	0.9588	0.8803	0.8885	0.8349	0.9069	0.9169	0.4743	0.6645
0.0571	1.29	6000	0.0551	0.8351	0.8934	0.9810	0.9957	0.9831	0.9652	0.9693	0.8562	0.9593	0.9569	0.5959	0.7586	0.9910	0.9579	0.8842	0.8879	0.8188	0.9084	0.9155	0.4774	0.6749
0.0778	1.4	6500	0.0537	0.8430	0.8994	0.9818	0.9948	0.9839	0.9872	0.9921	0.8702	0.9587	0.9635	0.5790	0.7656	0.9911	0.9579	0.9044	0.9093	0.8458	0.9060	0.9157	0.4760	0.6808
0.0392	1.51	7000	0.0491	0.8503	0.9069	0.9830	0.9954	0.9823	0.9645	0.9666	0.9205	0.9534	0.9599	0.6214	0.7984	0.9916	0.9607	0.9123	0.9139	0.8755	0.9072	0.9180	0.4907	0.6830
0.0376	1.62	7500	0.0514	0.8442	0.9010	0.9819	0.9954	0.9832	0.9652	0.9660	0.8850	0.9525	0.9598	0.6257	0.7762	0.9914	0.9586	0.8944	0.9053	0.8355	0.9104	0.9215	0.4965	0.6838
0.0391	1.73	8000	0.0492	0.8422	0.8993	0.9819	0.9958	0.9836	0.9641	0.9671	0.8692	0.9561	0.9661	0.6159	0.7756	0.9916	0.9596	0.8882	0.8930	0.8338	0.9103	0.9189	0.4982	0.6860
0.0446	1.83	8500	0.0491	0.8515	0.9079	0.9829	0.9960	0.9836	0.9890	0.9913	0.8770	0.9505	0.9631	0.6458	0.7751	0.9916	0.9603	0.9114	0.9161	0.8559	0.9100	0.9217	0.5096	0.6867
0.041	1.94	9000	0.0482	0.8464	0.8978	0.9825	0.9958	0.9848	0.9619	0.9668	0.8822	0.9569	0.9659	0.5961	0.7703	0.9916	0.9602	0.8958	0.9018	0.8438	0.9148	0.9231	0.4966	0.6899
0.0744	2.05	9500	0.0474	0.8523	0.9018	0.9834	0.9961	0.9840	0.9598	0.9633	0.9195	0.9471	0.9644	0.6055	0.7766	0.9919	0.9619	0.9095	0.9125	0.8697	0.9113	0.9238	0.5010	0.6889
0.0433	2.16	10000	0.0471	0.8581	0.9103	0.9842	0.9951	0.9843	0.9617	0.9646	0.9416	0.9549	0.9718	0.6305	0.7879	0.9915	0.9644	0.9100	0.9155	0.8976	0.9145	0.9245	0.5127	0.6920
0.0412	2.26	10500	0.0468	0.8574	0.9042	0.9835	0.9956	0.9848	0.9628	0.9669	0.9023	0.9615	0.9677	0.6115	0.7847	0.9918	0.9601	0.9248	0.9286	0.8656	0.9177	0.9245	0.5073	0.6964
0.0489	2.37	11000	0.0496	0.8511	0.9029	0.9832	0.9956	0.9858	0.9905	0.9948	0.8694	0.9574	0.9654	0.5748	0.7926	0.9921	0.9604	0.9066	0.9086	0.8615	0.9167	0.9228	0.4913	0.7004
0.0388	2.48	11500	0.0450	0.8594	0.9036	0.9849	0.9957	0.9857	0.9621	0.9648	0.9620	0.9493	0.9604	0.57

📄 License

This model is released 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