Space object detection improves with hybrid convolutional transformer system
TriCCOT: Tri-part Convolutional Conformal Transformer for Onboard Space Object Detection
Computer Vision and Pattern RecognitionArtificial Intelligence
Summary
Detecting objects from space can be tough because onboard computers have limited power and the images they get are often blurry or noisy. The researchers created a new system called TriCCOT that mixes two types of machine learning models: convolutional networks that quickly find likely objects, and a special transformer-based classifier that works well on limited hardware. Their system also uses a method to estimate uncertainty in object locations for better reliability. Tests showed that TriCCOT is good at finding objects even when images are unclear and can run on space-ready chips without extra modifications.
Convolutional neural networksTransformersObject detectionOnboard processingFPGA acceleratorsConformal predictionAttention mechanismsRegion proposal network
Authors
Adrien Dorise, Marjorie Bellizzi, Julia Cohen, Stéphane May
Abstract
Onboard object detection in Earth observation is constrained by limited computational resources and the absence of fully corrected imagery. While convolutional detectors are hardware-efficient, they often struggle to extract robust representations from raw and noisy data. Conversely, transformer-based models provide stronger global reasoning capabilities but remain difficult to deploy on FPGA accelerators due to quadratic attention complexity and non-compatible operations. We introduce TriCCOT, a tri-part architecture for robust and deployable onboard object detection. TriCCOT combines a convolutional region proposal network, a conformal prediction stage, and Aper-GATES, our hardware-friendly attention-based classifier. The region proposal network generates candidate bounding boxes, which are subsequently enlarged via conformal prediction, providing a distribution-free probabilistic coverage guarantee. The resulting crops are processed by Aper-GATES, which reformulates self-attention through convolutional projections, global channel statistics, and hardware-friendly gating operations, avoiding standard transformer operations that are poorly suited to CNN-oriented accelerators. Experiments on the DIOR and VDVRaw datasets demonstrate competitive detection performance and improved robustness to spatial blur and signal-dependent noise when compared to FPGA-compatible architectures. Finally, we report full deployment on a Xilinx Versal VCK190 FPGA without modifying the underlying DPU architecture, enabling unified CNN-Transformer inference for spaceborne embedded applications.