Segmentation-based Detection for Efficient Multi-Task Spacecraft Perception
Researchers have developed a novel segmentation-based detection method for multi-task spacecraft perception, addressing challenges like limited annotated data and difficult visual conditions. Their compact architecture, featuring a MobileNetV3 encoder and a U-Net-style decoder, achieved strong performance in classification, detection, and segmentation tasks. This approach ranked second in the SPARK 2026 Challenge, demonstrating the effectiveness of lightweight models for practical onboard space vision systems. AI