Robust Character Labeling in Movie Videos: Data Resources and Self-supervised Feature Adaptation
This work addresses the challenge of automatic character labeling in videos for media analysis, though it is incremental with new datasets and adaptation methods.
The paper tackles the problem of robust face clustering in movies by addressing the lack of domain-specific datasets and adapting face embeddings to long-form content, resulting in performance comparable to state-of-the-art methods on downstream tasks.
Robust face clustering is a vital step in enabling computational understanding of visual character portrayal in media. Face clustering for long-form content is challenging because of variations in appearance and lack of supporting large-scale labeled data. Our work in this paper focuses on two key aspects of this problem: the lack of domain-specific training or benchmark datasets, and adapting face embeddings learned on web images to long-form content, specifically movies. First, we present a dataset of over 169,000 face tracks curated from 240 Hollywood movies with weak labels on whether a pair of face tracks belong to the same or a different character. We propose an offline algorithm based on nearest-neighbor search in the embedding space to mine hard-examples from these tracks. We then investigate triplet-loss and multiview correlation-based methods for adapting face embeddings to hard-examples. Our experimental results highlight the usefulness of weakly labeled data for domain-specific feature adaptation. Overall, we find that multiview correlation-based adaptation yields more discriminative and robust face embeddings. Its performance on downstream face verification and clustering tasks is comparable to that of the state-of-the-art results in this domain. We also present the SAIL-Movie Character Benchmark corpus developed to augment existing benchmarks. It consists of racially diverse actors and provides face-quality labels for subsequent error analysis. We hope that the large-scale datasets developed in this work can further advance automatic character labeling in videos. All resources are available freely at https://sail.usc.edu/~ccmi/multiface.