Image Recognition of Prohibited Express Packages Based on YOLOv4
- Resource Type
- Conference
- Authors
- Cheng, Cheng; Wang, Xifu; Shen, Mengru; Cheng, Yan
- Source
- 2021 3rd International Academic Exchange Conference on Science and Technology Innovation (IAECST) Science and Technology Innovation (IAECST), 2021 3rd International Academic Exchange Conference on. :1413-1418 Dec, 2021
- Subject
- Communication, Networking and Broadcast Technologies
Computing and Processing
Engineering Profession
Transportation
Industries
Technological innovation
Image recognition
Transportation
Security
YOLOv4
prohibited express packages
deep learning
express parcel security
- Language
In the increasingly complex security environment of the express industry, the identification problem of prohibited items with small samples and multiple labels has not been properly solved. These prohibited items greatly increase the risk of safe transportation of express delivery and bring great security risks to the society. The key to solve this problem is to realize accurate and efficient identification of prohibited items with small samples and multiple labels. On this basis, the YOLOv4 algorithm model was used to realize the recognition of security check images of the delivered items, and the loss function value of the best model was trained to be 0.95 and the accuracy was 75%.