EdgeYOLO: An Edge-Real-Time Object Detector

An efficient, low-complexity, and anchor-free object detector based on the state-of-the-art YOLO framework is proposed in this paper, which can be implemented in real time on edge computing platforms. An enhanced data augmentation method is developed to effectively suppress overfitting during traini...

Full description

Saved in:
Bibliographic Details
Published inChinese Control Conference pp. 7507 - 7512
Main Authors Liu, Shihan, Zha, Junlin, Sun, Jian, Li, Zhuo, Wang, Gang
Format Conference Proceeding
LanguageEnglish
Published Technical Committee on Control Theory, Chinese Association of Automation 24.07.2023
Subjects
Anchor-free
Computational modeling
Detectors
edge-real-time
hybrid random loss
Image edge detection
object detector
Performance evaluation
Real-time systems
Source coding
Training
Online AccessGet full text

Cover

More Information
Summary:An efficient, low-complexity, and anchor-free object detector based on the state-of-the-art YOLO framework is proposed in this paper, which can be implemented in real time on edge computing platforms. An enhanced data augmentation method is developed to effectively suppress overfitting during training, and a hybrid random loss function is designed to improve the detection accuracy of small objects. Inspired by FCOS, a lighter and more efficient decoupled head is proposed, and its inference speed can be improved with little loss of precision. Our baseline model can reach the accuracy of 50.6% AP50:95 and 69.8% AP 50 in MS COC02017 dataset, 26.9% AP 50:95 and 45.4% AP 50 in VisDrone2019-DET dataset, and it meets real-time requirements (FPS230) on edge-computing device Nvidia Jetson AGX Xavier. And as is shown in Fig. 1, lighter models with less parameters designed for edge computing devices with lower computing power also show better performances. Our source code, hyper-parameters and model weights are all available at https://github.com/LSH9832/edgeyolo.
ISSN:1934-1768
DOI:10.23919/CCC58697.2023.10239786