This paper proposes a new Privacy-Preserving TFGC (Target Fine-grained Classification) Framework based on Federated Learning, named PRFL. The framework is designed to help clients learn global and local knowledge, improving the representation of private data. The PRFL also minimizes communication overhead, boosts efficiency, and ensures satisfactory performance, even under resource-scarce conditions. The framework’s effectiveness is demonstrated using four public datasets. This work is significant in the field of remote sensing, particularly in the context of data privacy and efficiency.
Publication date: 3 Jan 2024
Project Page: https://arxiv.org/abs/2401.01493
Paper: https://arxiv.org/pdf/2401.01493
