To address design plagiarism and insecure copyright management in the fashion industry, this paper proposes a unified framework that synergizes deep learning with blockchain cryptography. Our framework utilizes neural networks to provide AI-assisted evidence ranking for expert review. We introduce two core components. First, a Dual-Branch Attribute Embedding Network (DBAE-Net) enables precise similarity ranking of prior works by capturing both global style and fine-grained local details. This ranking assists human experts in identifying potential plagiarism. Second, we design a Traceable Verifiable Scheme (TVS) using bilinear pairings that provides an unforgeable on-chain proof of ownership. The TVS balances signer anonymity against external parties with robust, authority-led traceability. We formally prove the TVS’s security and empirically validate the framework’s effectiveness, offering a novel, pragmatic solution for streamlining copyright verification in the digital fashion ecosystem.