Fine-grained identification of camera devices based on inherent features

Area of Science:

• Computer Vision
• Device Identification
• Cybersecurity

Background:

• Camera devices are increasingly prevalent in smart homes, cities, and enterprises.
• Accurate device identification is crucial for security and understanding device characteristics.
• Existing methods struggle with fine-grained device type differentiation.

Purpose of the Study:

• To develop a fine-grained device identification method for camera devices.
• To leverage inherent device features for accurate classification.
• To improve device security through precise identification.

Main Methods:

• Feature selection based on coverage and differences of inherent features.
• Classification of features by representation and establishment of a feature similarity calculation strategy (FSCS).
• Determination of feature weights using feature entropy and development of a device similarity model.

Main Results:

• The proposed model successfully identifies fine-grained camera device types.
• The method demonstrates robustness, achieving over 80% accuracy with up to 50% missing inherent feature values.
• Evaluation on Dahua and Hikvision camera devices validates the method's effectiveness.

Conclusions:

• The developed device similarity model accurately identifies fine-grained camera device types.
• The method offers a reliable solution for device identification, even with incomplete data.
• This contributes to enhanced security and management of networked camera devices.