The problem ZKFaceRecovery solves

ZKFaceRecovery empowers users to securely recover their wallets using their facial features and personal information without exposing these sensitive details. By leveraging facial recognition, fuzzy commitment for secure key generation, and zero-knowledge proofs for privacy preservation, it provides a robust and private way for users to regain access to their digital wallets.
Key components are:
Facial Recognition as a Key: It uses facial features to generate an embedding, acting as a key to a cryptographic commitment.
Fuzzy Commitment: Combines the feature vector with an error-encoded secret to create a commitment.
Error Correction: Utilizes Reed-Solomon algorithm for error correction within the commitment, enhancing reliability.
Zero-Knowledge Proofs: Ensures privacy by verifying identity without exposing the actual facial data.
Blockchain: Employs blockchain for securely storing commitment data, not the facial vector.

This approach offers robust security and privacy for digital wallet recovery and similar applications.

Challenges I ran into

Building ZKFaceRecovery involved overcoming several challenges:

Understanding Complex Algorithms: Grasping the nuances of error correction algorithms and fuzzy commitment schemes was technically demanding.
Circom Circuit Optimization: Writing and optimizing the Circom circuit to handle complex operations while ensuring privacy posed a significant challenge.
Balancing Security and Usability: Ensuring robust security without compromising user experience required careful design and testing.
Integrating and e2e working: Seamlessly integrating machine learning, blockchain, and cryptographic techniques was quite difficult.