Leveraging Self-Supervision for Cross-Domain Crowd Counting

TL;DR

This paper introduces a novel cross-domain crowd counting method that combines synthetic and unlabeled real images using self-supervision, improving generalization without extra inference costs.

Contribution

It proposes a self-supervised training approach that leverages unlabeled real images and synthetic data, enhancing cross-domain crowd counting performance.

Findings

Outperforms state-of-the-art cross-domain crowd counting methods

Uses perspective-aware feature learning and uncertainty prediction

No additional inference computation required

Abstract

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.