Paper detail

Red Carpet to Fight Club: Partially-supervised Domain Transfer for Face Recognition in Violent Videos

In many real-world problems, there is typically a large discrepancy between the characteristics of data used in training versus deployment. A prime example is the analysis of aggression videos: in a criminal incidence, typically suspects need to be identified based on their clean portrait-like photos, instead of their prior video recordings. This results in three major challenges; large domain discrepancy between violence videos and ID-photos, the lack of video examples for most individuals and limited training data availability. To mimic such scenarios, we formulate a realistic domain-transfer problem, where the goal is to transfer the recognition model trained on clean posed images to the target domain of violent videos, where training videos are available only for a subset of subjects. To this end, we introduce the WildestFaces dataset, tailored to study cross-domain recognition under a variety of adverse conditions. We divide the task of transferring a recognition model from the domain of clean images to the violent videos into two sub-problems and tackle them using (i) stacked affine-transforms for classifier-transfer, (ii) attention-driven pooling for temporal-adaptation. We additionally formulate a self-attention based model for domain-transfer. We establish a rigorous evaluation protocol for this clean-to-violent recognition task, and present a detailed analysis of the proposed dataset and the methods. Our experiments highlight the unique challenges introduced by the WildestFaces dataset and the advantages of the proposed approach.