Provenance and Attribution

Images are copied, cropped, recombined and regenerated as they move across the web. We trace these relationships back to their sources, reconstructing how a set of related images was derived from one another and attributing synthetic content to the identities and models behind it. Our work spans transformation-aware embeddings, large-scale provenance retrieval, and identity-based provenance for generated faces.

Publications

SynthProv: Interpretable Framework for Profiling Identity Leakage

Jaisidh Singh, Harshil Bhatia, Mayank Vatsa, Richa Singh, Aparna Bharati

2024 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)  ·  03 Jan 2024  ·  doi:10.1109/wacv57701.2024.00468

IdProv: Identity-Based Provenance for Synthetic Image Generation (Student Abstract)

Harshil Bhatia, Jaisidh Singh, Gaurav Sangwan, Aparna Bharati, Richa Singh, Mayank Vatsa

Proceedings of the AAAI Conference on Artificial Intelligence  ·  26 Jun 2023  ·  doi:10.1609/aaai.v37i13.26942

Image Provenance Analysis

Daniel Moreira, William Theisen, Walter Scheirer, Aparna Bharati, Joel Brogan, Anderson Rocha

Advances in Computer Vision and Pattern Recognition  ·  01 Jan 2022  ·  doi:10.1007/978-981-16-7621-5_15

Transformation-Aware Embeddings for Image Provenance

Aparna Bharati, Daniel Moreira, Patrick J. Flynn, Anderson de Rezende Rocha, Kevin W. Bowyer, Walter J. Scheirer

IEEE Transactions on Information Forensics and Security  ·  01 Jan 2021  ·  doi:10.1109/TIFS.2021.3050061

Fast Local Spatial Verification for Feature-Agnostic Large-Scale Image Retrieval

Joel Brogan, Aparna Bharati, Daniel Moreira, Anderson Rocha, Kevin W. Bowyer, Patrick J. Flynn, Walter J. Scheirer

IEEE Transactions on Image Processing  ·  01 Jan 2021  ·  doi:10.1109/TIP.2021.3097175