Leveraging Self-Supervision for Cross-Domain Crowd Counting
This addresses the challenge of deploying crowd counting models in emergencies where data annotation is costly or slow, though it is incremental as it builds on existing cross-domain approaches.
The paper tackles the problem of poor generalization in cross-domain crowd counting when using synthetic data for training by incorporating self-supervised learning with unlabeled real images, resulting in an algorithm that consistently outperforms state-of-the-art methods without extra inference cost.
State-of-the-art methods for counting people in crowded scenes rely on deep networks to estimate crowd density. While effective, these data-driven approaches rely on large amount of data annotation to achieve good performance, which stops these models from being deployed in emergencies during which data annotation is either too costly or cannot be obtained fast enough. One popular solution is to use synthetic data for training. Unfortunately, due to domain shift, the resulting models generalize poorly on real imagery. We remedy this shortcoming by training with both synthetic images, along with their associated labels, and unlabeled real images. To this end, we force our network to learn perspective-aware features by training it to recognize upside-down real images from regular ones and incorporate into it the ability to predict its own uncertainty so that it can generate useful pseudo labels for fine-tuning purposes. This yields an algorithm that consistently outperforms state-of-the-art cross-domain crowd counting ones without any extra computation at inference time.