Domain penalisation for improved Out-of-Distribution Generalisation

TL;DR

This paper introduces a domain penalisation framework for object detection that dynamically adjusts focus on source domains to improve out-of-distribution generalisation, showing consistent performance gains on benchmark datasets.

Contribution

It proposes a novel domain penalisation method for object detection that balances training across multiple source domains to enhance OOD performance.

Findings

Improves FasterRCNN OOD accuracy by 0.3% and 0.5%.

Enhances FCOS detector OOD performance by over 1.3%.

Demonstrates effectiveness on GWHD 2021 dataset.

Abstract

In the field of object detection, domain generalisation (DG) aims to ensure robust performance across diverse and unseen target domains by learning the robust domain-invariant features corresponding to the objects of interest across multiple source domains. While there are many approaches established for performing DG for the task of classification, there has been a very little focus on object detection. In this paper, we propose a domain penalisation (DP) framework for the task of object detection, where the data is assumed to be sampled from multiple source domains and tested on completely unseen test domains. We assign penalisation weights to each domain, with the values updated based on the detection networks performance on the respective source domains. By prioritising the domains that needs more attention, our approach effectively balances the training process. We evaluate our solution on the GWHD 2021 dataset, a component of the WiLDS benchmark and we compare against ERM and GroupDRO as these are primarily loss function based. Our extensive experimental results reveals that the proposed approach improves the accuracy by 0.3 percent and 0.5 percent on validation and test out-of-distribution (OOD) sets, respectively for FasterRCNN. We also compare the performance of our approach on FCOS detector and show that our approach improves the baseline OOD performance over the existing approaches by 1.3 percent and 1.4 percent on validation and test sets, respectively. This study underscores the potential of performance based domain penalisation in enhancing the generalisation ability of object detection models across diverse environments.