Randomize to Generalize: Domain Randomization for Runway FOD Detection

TL;DR

This paper introduces a novel two-stage synthetic data augmentation method called SRIA to improve the generalization of tiny object detection models, especially for foreign object debris detection in out-of-distribution scenarios.

Contribution

The paper proposes SRIA, a two-stage synthetic augmentation technique that enhances model generalization on 2D datasets without relying on 3D rendering, demonstrated on FOD detection.

Findings

Detection accuracy improved from 41% to 92% on OOD test set.

Synthetic data generation using SRIA significantly boosts out-of-distribution detection performance.

Method outperforms several state-of-the-art models on FOD dataset.

Abstract

Tiny Object Detection is challenging due to small size, low resolution, occlusion, background clutter, lighting conditions and small object-to-image ratio. Further, object detection methodologies often make underlying assumption that both training and testing data remain congruent. However, this presumption often leads to decline in performance when model is applied to out-of-domain(unseen) data. Techniques like synthetic image generation are employed to improve model performance by leveraging variations in input data. Such an approach typically presumes access to 3D-rendered datasets. In contrast, we propose a novel two-stage methodology Synthetic Randomized Image Augmentation (SRIA), carefully devised to enhance generalization capabilities of models encountering 2D datasets, particularly with lower resolution which is more practical in real-world scenarios. The first stage employs a weakly supervised technique to generate pixel-level segmentation masks. Subsequently, the second stage generates a batch-wise synthesis of artificial images, carefully designed with an array of diverse augmentations. The efficacy of proposed technique is illustrated on challenging foreign object debris (FOD) detection. We compare our results with several SOTA models including CenterNet, SSD, YOLOv3, YOLOv4, YOLOv5, and Outer Vit on a publicly available FOD-A dataset. We also construct an out-of-distribution test set encompassing 800 annotated images featuring a corpus of ten common categories. Notably, by harnessing merely 1.81% of objects from source training data and amalgamating with 29 runway background images, we generate 2227 synthetic images. Subsequent model retraining via transfer learning, utilizing enriched dataset generated by domain randomization, demonstrates significant improvement in detection accuracy. We report that detection accuracy improved from an initial 41% to 92% for OOD test set.