facebook-detr-resnet-50-finetuned-10k-cppe5 Open Source Model - Free Deployment for Precise Detection of Medical Protective Equipment

Facebook Detr Resnet 50 Finetuned 10k Cppe5 With Augs

Developed by qubvel-hf

Object detection model fine-tuned on the cppe-5 dataset based on facebook/detr-resnet-50, specializing in medical PPE detection

Object Detection

Transformers

Open Source License:Apache-2.0 #Medical PPE Detection #Object Detection Fine-tuning #Multi-scale Object Recognition

Downloads 19

Release Time : 4/30/2024

Model Overview

This model is a ResNet-50 variant based on the DETR (Detection Transformer) architecture, specifically fine-tuned for medical personal protective equipment (PPE) detection tasks.

Model Features

Medical PPE Detection

Detection capability specifically optimized for medical protective equipment such as gowns, face shields, gloves, goggles, and masks

Multi-scale Detection Capability

Capable of detecting objects of various sizes, including small, medium, and large protective equipment

Data Augmentation Training

Utilized data augmentation techniques during training to enhance model generalization

Model Capabilities

Medical PPE Detection

Multi-category Object Recognition

Object Localization in Images

Small Object Detection

Use Cases

Medical Safety

Hospital PPE Compliance Detection

Automatically detect whether medical staff are wearing complete protective equipment

Can identify five main types of PPE with an average precision (mAP) of 0.2873

Epidemic Prevention Monitoring

Monitor PPE usage in public places

Mask detection achieves mAP of 0.2871 with recall@100 of 0.3907

Industrial Safety

Factory Safety Equipment Inspection

Automatically detect whether workers are wearing necessary protective equipment

Protective gown detection achieves mAP of 0.5461 with recall@100 of 0.6932

🚀 facebook-detr-resnet-50-finetuned-10k-cppe5-with-augs

This model is a fine - tuned version of [facebook/detr - resnet - 50](https://huggingface.co/facebook/detr - resnet - 50) on the cppe - 5 dataset, which can be used for object detection tasks.

[](https://wandb.ai/qubvel - hf - co/transformers - detection - model - finetuning - cppe5/runs/u73br2l9)

🚀 Quick Start

This model is a fine - tuned version of [facebook/detr - resnet - 50](https://huggingface.co/facebook/detr - resnet - 50) on the cppe - 5 dataset. It achieves the following results on the evaluation set:

Loss: 1.2371
Map: 0.2873
Map 50: 0.5678
Map 75: 0.2501
Map Small: 0.126
Map Medium: 0.2327
Map Large: 0.4873
Mar 1: 0.2843
Mar 10: 0.4643
Mar 100: 0.4762
Mar Small: 0.2338
Mar Medium: 0.4167
Mar Large: 0.7114
Map Coverall: 0.5461
Mar 100 Coverall: 0.6932
Map Face Shield: 0.2167
Mar 100 Face Shield: 0.4785
Map Gloves: 0.2135
Mar 100 Gloves: 0.4094
Map Goggles: 0.173
Mar 100 Goggles: 0.4092
Map Mask: 0.2871
Mar 100 Mask: 0.3907

🔧 Technical Details

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 5e - 05
train_batch_size: 8
eval_batch_size: 8
seed: 1337
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
num_epochs: 100.0
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 Coverall	Mar 100 Coverall	Map Face Shield	Mar 100 Face Shield	Map Gloves	Mar 100 Gloves	Map Goggles	Mar 100 Goggles	Map Mask	Mar 100 Mask
2.6324	1.0	107	2.3696	0.0355	0.0772	0.0305	0.0094	0.0044	0.0388	0.0606	0.1313	0.1643	0.0314	0.0921	0.1907	0.1559	0.536	0.0	0.0	0.0064	0.1433	0.0	0.0	0.0151	0.1422
2.0851	2.0	214	2.1468	0.0518	0.1164	0.0389	0.0106	0.0151	0.0683	0.0809	0.1628	0.1881	0.023	0.1069	0.2515	0.2068	0.5842	0.0	0.0	0.0174	0.167	0.0	0.0	0.035	0.1893
1.9849	3.0	321	2.0694	0.0592	0.1436	0.0377	0.0187	0.023	0.0609	0.0796	0.1666	0.191	0.0527	0.1032	0.2411	0.2119	0.5486	0.0	0.0	0.0226	0.1799	0.0	0.0	0.0615	0.2262
1.8364	4.0	428	2.0431	0.0715	0.1666	0.0625	0.0031	0.0232	0.09	0.0775	0.1641	0.183	0.0083	0.1006	0.2708	0.2923	0.5604	0.0012	0.0177	0.0203	0.1612	0.0	0.0	0.0437	0.1756
1.7433	5.0	535	1.9001	0.0938	0.2115	0.0758	0.0229	0.0473	0.1031	0.1066	0.216	0.2311	0.0451	0.1423	0.3186	0.3168	0.5847	0.0064	0.0937	0.05	0.2277	0.0	0.0	0.0959	0.2493
1.7113	6.0	642	1.8483	0.1007	0.2225	0.0773	0.0203	0.0593	0.1341	0.1109	0.2213	0.2463	0.0646	0.1711	0.3303	0.3462	0.5914	0.0132	0.0987	0.0451	0.2571	0.0002	0.0062	0.0987	0.2782
1.6518	7.0	749	1.8376	0.1055	0.2347	0.0807	0.0199	0.0596	0.1405	0.1127	0.2364	0.2598	0.0778	0.1949	0.3276	0.3693	0.5869	0.0168	0.1367	0.0417	0.2397	0.0046	0.0523	0.0953	0.2831
1.6048	8.0	856	1.8041	0.1098	0.2576	0.0857	0.0155	0.0772	0.1578	0.1284	0.2423	0.2708	0.0602	0.2057	0.3631	0.3512	0.6149	0.033	0.1342	0.0402	0.2491	0.004	0.0862	0.1204	0.2698
1.6026	9.0	963	1.7173	0.1243	0.2773	0.1041	0.0415	0.0688	0.1767	0.1426	0.2852	0.3101	0.0857	0.2275	0.4249	0.3889	0.6378	0.0347	0.2253	0.0547	0.271	0.0159	0.1308	0.1274	0.2858
1.5298	10.0	1070	1.7915	0.1154	0.2763	0.0804	0.018	0.0789	0.1762	0.1447	0.2629	0.2913	0.0774	0.2191	0.3935	0.3615	0.618	0.0415	0.2127	0.0351	0.2603	0.0076	0.0846	0.1315	0.2809
1.5114	11.0	1177	1.7964	0.1232	0.2755	0.0915	0.0396	0.0849	0.1728	0.144	0.2876	0.3007	0.0868	0.2161	0.4535	0.3588	0.5923	0.035	0.2291	0.057	0.2241	0.0195	0.1877	0.1455	0.2702
1.5283	12.0	1284	1.7017	0.1444	0.3075	0.1223	0.0215	0.0985	0.2396	0.1803	0.3244	0.3456	0.0531	0.282	0.5554	0.4122	0.632	0.0666	0.3215	0.0618	0.2625	0.0086	0.2046	0.1727	0.3076
1.5034	13.0	1391	1.6344	0.1538	0.3418	0.1226	0.0423	0.1165	0.2381	0.1871	0.337	0.3578	0.1195	0.2901	0.5214	0.4204	0.6284	0.0705	0.319	0.0634	0.2884	0.0191	0.2277	0.1959	0.3253
1.445	14.0	1498	1.5685	0.167	0.3626	0.15	0.0561	0.1266	0.2715	0.1998	0.357	0.3701	0.1068	0.2922	0.5693	0.4227	0.6342	0.0873	0.3797	0.0744	0.2951	0.0441	0.2108	0.2065	0.3307
1.3998	15.0	1605	1.5679	0.1645	0.3515	0.1383	0.0269	0.117	0.2592	0.1992	0.3295	0.3539	0.0645	0.2751	0.5533	0.4535	0.6703	0.0887	0.3494	0.0866	0.3116	0.0276	0.1554	0.1659	0.2827
1.398	16.0	1712	1.6191	0.1578	0.3459	0.1205	0.0623	0.0967	0.258	0.1729	0.3255	0.3386	0.1106	0.2686	0.5066	0.4363	0.6	0.0971	0.3089	0.0703	0.2839	0.0258	0.2	0.1593	0.3004
1.369	17.0	1819	1.6099	0.1654	0.3517	0.1371	0.0341	0.099	0.2543	0.1852	0.3085	0.3232	0.0587	0.2331	0.5126	0.4663	0.6329	0.0784	0.2785	0.0761	0.2906	0.019	0.12	0.187	0.2942
1.3264	18.0	1926	1.5610	0.1784	0.3752	0.1594	0.036	0.1249	0.3082	0.2061	0.3388	0.3561	0.0959	0.2702	0.5669	0.5	0.6707	0.0768	0.2937	0.0899	0.2933	0.0291	0.2108	0.1962	0.312
1.3513	19.0	2033	1.4462	0.1864	0.3933	0.154	0.0559	0.1364	0.2738	0.2135	0.3767	0.3931	0.1216	0.3178	0.5744	0.4834	0.677	0.0976	0.3595	0.1022	0.3438	0.0248	0.2554	0.224	0.3298
1.3034	20.0	2140	1.4527	0.1965	0.412	0.1713	0.051	0.1463	0.2967	0.2087	0.3643	0.3805	0.1422	0.2931	0.5833	0.4982	0.6495	0.1144	0.362	0.0997	0.3263	0.0497	0.2277	0.2204	0.3369
1.2823	21.0	2247	1.4872	0.1835	0.3972	0.1556	0.0596	0.1308	0.282	0.2148	0.3924	0.4147	0.1236	0.3657	0.5903	0.4622	0.6455	0.0727	0.3873	0.1195	0.3558	0.0544	0.3415	0.2084	0.3431
1.3099	22.0	2354	1.4723	0.1903	0.407	0.1632	0.0594	0.1343	0.2937	0.2158	0.3817	0.396	0.1357	0.3222	0.5737	0.4719	0.6486	0.1089	0.3722	0.1153	0.3161	0.0404	0.3092	0.2149	0.3338
1.2691	23.0	2461	1.4648	0.1897	0.403	0.1626	0.0598	0.1476	0.2933	0.2129	0.3676	0.3851	0.1154	0.3247	0.592	0.4836	0.6658	0.0711	0.3025	0.1217	0.3679	0.0594	0.2615	0.2127	0.328
1.3018	24.0	2568	1.5593	0.174	0.3964	0.1358	0.0568	0.1344	0.2776	0.2076	0.3571	0.3754	0.1412	0.2946	0.5732	0.4208	0.6347	0.0782	0.2987	0.1125	0.3304	0.0444	0.2831	0.2141	0.3302
1.2744	25.0	2675	1.4785	0.1863	0.3983	0.1556	0.0706	0.1314	0.3032	0.221	0.3784	0.3988	0.1453	0.3331	0.5824	0.4831	0.6716	0.0934	0.3392	0.1155	0.3647	0.0612	0.3277	0.1783	0.2907
1.3089	26.0	2782	1.4503	0.1969	0.4129	0.166	0.0904	0.1342	0.3274	0.2303	0.3907	0.4057	0.154	0.3252	0.6182	0.4773	0.6653	0.109	0.4114	0.1176	0.3504	0.0758	0.2877	0.2046	0.3138
1.2391	27.0	2889	1.4961	0.1992	0.4097	0.1714	0.067	0.1354	0.3227	0.2308	0.3803	0.3971	0.1333	0.3179	0.623	0.4826	0.6473	0.114	0.3329	0.1169	0.3621	0.0538	0.2862	0.2286	0.3569
1.2156	28.0	2996	1.4349	0.1926	0.4121	0.1612	0.0655	0.1425	0.2933	0.2279	0.4047	0.425	0.1521	0.36	0.6217	0.4698	0.6797	0.0974	0.3924	0.1097	0.3402	0.061	0.3677	0.2253	0.3449
1.2131	29.0	3103	1.4869	0.2002	0.4106	0.1706	0.0662	0.1516	0.3142	0.2342	0.3959	0.4175	0.1389	0.3525	0.6269	0.4846	0.6514	0.0845	0.3899	0.1196	0.3701	0.0699	0.3277	0.2425	0.3484
1.2368	30.0	3210	1.3779	0.218	0.4403	0.185	0.084	0.169	0.3567	0.241	0.4115	0.4326	0.148	0.3738	0.6216	0.5063	0.6815	0.122	0.4342	0.1497	0.3969	0.066	0.3031	0.2462	0.3471
1.1998	31.0	3317	1.4472	0.198	0.4299	0.16	0.0595	0.1492	0.3264	0.2239	0.3883

📄 License

This model is licensed under the Apache 2.0 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