Learning Cascaded Detection Tasks with Weakly-Supervised Domain Adaptation

TL;DR

This paper introduces a weakly supervised domain adaptation method for cascaded detection tasks in autonomous driving, effectively reducing annotation costs and handling domain shifts between synthetic and real data.

Contribution

It proposes leveraging 2D bounding boxes as weak labels and class-wise feature alignment to improve domain adaptation in cascaded detection tasks.

Findings

Competitive with fully supervised methods

Outperforms unsupervised domain adaptation approaches

Effective in reducing annotation requirements

Abstract

In order to handle the challenges of autonomous driving, deep learning has proven to be crucial in tackling increasingly complex tasks, such as 3D detection or instance segmentation. State-of-the-art approaches for image-based detection tasks tackle this complexity by operating in a cascaded fashion: they first extract a 2D bounding box based on which additional attributes, e.g. instance masks, are inferred. While these methods perform well, a key challenge remains the lack of accurate and cheap annotations for the growing variety of tasks. Synthetic data presents a promising solution but, despite the effort in domain adaptation research, the gap between synthetic and real data remains an open problem. In this work, we propose a weakly supervised domain adaptation setting which exploits the structure of cascaded detection tasks. In particular, we learn to infer the attributes solely from the source domain while leveraging 2D bounding boxes as weak labels in both domains to explain the domain shift. We further encourage domain-invariant features through class-wise feature alignment using ground-truth class information, which is not available in the unsupervised setting. As our experiments demonstrate, the approach is competitive with fully supervised settings while outperforming unsupervised adaptation approaches by a large margin.