Domain Adaptive Object Detection for Space Applications with Real-Time Constraints

2025-09-22

AI Summary

To address the significant performance degradation of synthetic-data-trained models on real-world space target detection due to domain shift, this paper proposes a lightweight supervised domain adaptation framework. The method integrates domain-invariant feature learning, a CNN-based domain discriminator, and a domain-agnostic regression head, augmented by semi-supervised learning and invariant risk minimization (IRM). Leveraging only 250 labeled real-world images, it achieves efficient cross-domain adaptation on lightweight detectors—MobileNet-SSD and ResNet-FCOS. On the SPEED+ and SPARK benchmarks, the approach improves mean average precision (mAP) by up to 20 percentage points, substantially narrowing the domain gap while satisfying onboard real-time inference constraints. This work establishes a deployable domain adaptation paradigm for resource-constrained spaceborne intelligent perception systems.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Adversarial Attacks & Robustness
• Application Domains: Mobility, Driving & Flight

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Object detection is essential in space applications targeting Space Domain Awareness and also applications involving relative navigation scenarios. Current deep learning models for Object Detection in space applications are often trained on synthetic data from simulators, however, the model performance drops significantly on real-world data due to the domain gap. However, domain adaptive object detection is an overlooked problem in the community. In this work, we first show the importance of domain adaptation and then explore Supervised Domain Adaptation (SDA) to reduce this gap using minimal labeled real data. We build on a recent semi-supervised adaptation method and tailor it for object detection. Our approach combines domain-invariant feature learning with a CNN-based domain discriminator and invariant risk minimization using a domain-independent regression head. To meet real-time deployment needs, we test our method on a lightweight Single Shot Multibox Detector (SSD) with MobileNet backbone and on the more advanced Fully Convolutional One-Stage object detector (FCOS) with ResNet-50 backbone. We evaluated on two space datasets, SPEED+ and SPARK. The results show up to 20-point improvements in average precision (AP) with just 250 labeled real images.

Problem

• Addressing domain gap between synthetic and real space object detection data
• Improving object detection accuracy for space applications with real-time constraints
• Developing domain adaptation methods using minimal labeled real space imagery

Innovation

• Supervised Domain Adaptation with minimal labeled data
• Domain-invariant feature learning with CNN discriminator
• Real-time lightweight detectors SSD and FCOS tested