Rtdetr R50 Cppe5 Finetune

Developed by qubvel-hf

Object detection model based on RT-DETR architecture, fine-tuned on CPPE-5 dataset with pre-training on COCO and Objects365 datasets

Object Detection

Transformers

#Object Detection #Medical Protective Equipment Recognition #Multi-class Detection

Downloads 207

Release Time : 5/29/2024

Model Overview

This model is a ResNet50 variant of RT-DETR (Real-Time Detection Transformer), designed for real-time object detection tasks, excelling in personal protective equipment (PPE) detection scenarios

Model Features

Real-time detection capability

Optimized based on Transformer architecture to achieve efficient real-time object detection

Multi-scale feature fusion

Effectively detects objects of different sizes, with particularly good recognition capability for small objects

PPE-specific optimization

Specially optimized for personal protective equipment (protective clothing, face shields, etc.) detection tasks

Model Capabilities

Real-time object detection

Multi-class object recognition

Small object detection

Medium-to-large object detection

Use Cases

Industrial safety

Personal protective equipment compliance detection

Detects whether workers are correctly wearing safety equipment such as protective clothing and masks

Protective clothing detection mAP reaches 0.6271, recall rate 0.8308

Medical protection

Medical protective monitoring

Monitors the wearing status of protective equipment by medical staff

Mask detection mAP reaches 0.5354, goggles detection mAP reaches 0.425

base_model: PekingU/rtdetr_r50vd_coco_o365 tags:

generated_from_trainer model-index:
name: rtdetr-r50-cppe5-finetune results: []

rtdetr-r50-cppe5-finetune

This model is a fine-tuned version of PekingU/rtdetr_r50vd_coco_o365 on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 9.7524
Map: 0.5298
Map 50: 0.7903
Map 75: 0.5632
Map Small: 0.5092
Map Medium: 0.4212
Map Large: 0.6655
Mar 1: 0.4001
Mar 10: 0.6526
Mar 100: 0.711
Mar Small: 0.6038
Mar Medium: 0.5835
Mar Large: 0.8378
Map Coverall: 0.6271
Mar 100 Coverall: 0.8308
Map Face Shield: 0.4839
Mar 100 Face Shield: 0.7706
Map Gloves: 0.5775
Mar 100 Gloves: 0.6492
Map Goggles: 0.425
Mar 100 Goggles: 0.6103
Map Mask: 0.5354
Mar 100 Mask: 0.6941

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: 5e-05
train_batch_size: 8
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 300
num_epochs: 10

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
No log	1.0	107	216.6647	0.0037	0.0089	0.0022	0.0032	0.0183	0.014	0.0242	0.1046	0.1966	0.0405	0.1831	0.4092	0.0056	0.2649	0.001	0.1962	0.0021	0.0719	0.0008	0.2215	0.0091	0.2284
No log	2.0	214	96.4364	0.0294	0.0559	0.0257	0.0169	0.0297	0.0299	0.0707	0.1835	0.298	0.0948	0.2203	0.4591	0.0888	0.5527	0.001	0.3203	0.021	0.1259	0.0014	0.2154	0.0346	0.2756
No log	3.0	321	28.5504	0.1576	0.294	0.1448	0.0752	0.0925	0.2629	0.1621	0.3534	0.4661	0.347	0.3964	0.6546	0.4399	0.6518	0.0021	0.3797	0.1282	0.3866	0.0045	0.4	0.2132	0.5124
No log	4.0	428	17.1997	0.2324	0.408	0.2295	0.1228	0.1816	0.3288	0.2317	0.4133	0.5	0.3527	0.4438	0.6543	0.5101	0.6396	0.0093	0.4671	0.1827	0.4513	0.1553	0.4062	0.3045	0.5356
117.1144	5.0	535	14.8812	0.2495	0.4498	0.2479	0.1261	0.1962	0.4086	0.253	0.4388	0.5189	0.3485	0.4683	0.7111	0.5078	0.6752	0.0291	0.5013	0.2265	0.4491	0.1715	0.4246	0.3129	0.5444
117.1144	6.0	642	13.5348	0.2572	0.4698	0.2541	0.1377	0.1905	0.424	0.2532	0.4315	0.4895	0.314	0.4481	0.6649	0.5166	0.6716	0.026	0.4873	0.2391	0.3754	0.1866	0.3754	0.3178	0.5378
117.1144	7.0	749	12.7545	0.2812	0.5035	0.2612	0.1618	0.2143	0.4653	0.2595	0.4568	0.496	0.3394	0.4438	0.6648	0.5152	0.6815	0.0918	0.4949	0.2504	0.3759	0.208	0.3954	0.3405	0.5324
117.1144	8.0	856	12.5330	0.2909	0.5328	0.2687	0.1568	0.2262	0.4868	0.2831	0.4625	0.5035	0.3209	0.4428	0.686	0.5059	0.6838	0.1762	0.5038	0.2528	0.3978	0.1905	0.4062	0.3289	0.5258
117.1144	9.0	963	12.2873	0.3023	0.5355	0.2927	0.1621	0.2502	0.494	0.2851	0.4696	0.5064	0.3301	0.452	0.6736	0.5276	0.6932	0.1696	0.4899	0.2633	0.4085	0.2249	0.4154	0.326	0.5249
16.4463	10.0	1070	12.2585	0.3095	0.5506	0.3029	0.1738	0.2405	0.4996	0.2901	0.4721	0.5105	0.3271	0.4558	0.6864	0.5196	0.6892	0.2225	0.5241	0.264	0.4022	0.2102	0.4077	0.3309	0.5293

Framework versions

Transformers 4.42.0.dev0
Pytorch 2.1.0+cu118
Datasets 2.19.1
Tokenizers 0.19.1

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