Selected Publications

Multi-view Classification Using Hybrid Fusion and Mutual Distillation
Our novel approach adapts off-the-shelf hybrid CNN/Transformer networks to achieve state-of-the-art performance classifying unstructured (e.g., scene identification) and cross-view (e.g., chest X-ray diagnosis) image collections.
Samuel Black and Richard Souvenir. In IEEE Winter Conference on Applications of Computer Vision (WACV), 2024.

Visualizing Paired Image Similarity in Transformer Networks
We introduce novel approach for spatially decomposing the similarity between a pair of images encoded with an image Transformer.
Samuel Black, Abby Stylianou, Robert Pless, and Richard Souvenir. In IEEE Winter Conference on Applications of Computer Vision (WACV), 2022.

The 2021 Hotel-ID to Combat Human Trafficking Competition Dataset
We develop a new dataset and Kaggle competition for fine-grained hotel room image classification.
Rashmi Kamath, Greg Rolwes, Samuel Black, and Abby Stylianou. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Fine-Grained Visual Categorization Workshop, 2021.

Evaluation of Inpainting and Augmentation for Censored Image Queries
This research evaluates the effectiveness of image inpainting and data augmentation methods in mitigating the impact of censoring on image classification and retrieval.
Journal: Samuel Black, Somayeh Keshavarz, and Richard Souvenir. In International Journal of Computer Vision (IJCV), pages 977-997, 2021.
Conference: Samuel Black, Somayeh Keshavarz, and Richard Souvenir. In IEEE Winter Conference on Applications of Computer Vision (WACV), 2020.