cs.CVNov 9, 2026

SpinSCI: Mask-Conditioned Spatial-Temporal Spin Transformer for Video Snapshot Compressive Imaging

Authors: Seungki Hong, Zhishuai Zhang, Taekwang Jang

Organizations: ETH AI Center, ETH Zurich, Switzerland

Abstract

Video snapshot compressive imaging (SCI) recovers multiple frames from one spatiotemporally coded measurement. This paper proposes SpinSCI, a mask-conditioned Spatial-Temporal Spin Transformer whose Mask-Conditioned Ising Refinement Block (MC-IRB) combines attention logits and sensing information to form signed spin couplings. Finite mean-field updates gate residual feature refinement. Physics-guided initialization and final data consistency incorporate the measurement model. Compared with HiSViT9_9, SpinSCI achieves average PSNR gains of 0.51 dB, 0.23 dB, and 0.04 dB on small-scale gray, mid-scale color, and large-scale color benchmarks, respectively, with 16.9% fewer parameters, 21.0-21.6% fewer FLOPs, 17.1-49.9% lower latency, and 62.3-63.3% lower peak memory. Relative to DU-9stg, average PSNR differs by at most 0.13 dB across these benchmarks, with 28.2-28.7% fewer FLOPs, 64.6-71.7% lower inference latency, and 30.1-35.0% lower peak memory. Qualitative comparisons on the real benchmark further show recovery of fine details and textures. Code and reference checkpoints are available at https://github.com/skethz/SpinSCI.

Figures & tables

Appendix figures & tables31 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. RobustSCI: Beyond Reconstruction to Restoration for Snapshot Compressive Imaging under Real-World Degradations

    Mar 8, 2026Hao Wang, Zhankuo Xu, Jiong Ni +3Image RestorationImage Corruption Robustness

  2. RayFormer: Modeling Inter- and Intra-Ray Similarity for NeRF-Based Video Snapshot Compressive Imaging

    Apr 30, 2026Yubo Dong, Danhua Liu, Anqi Li +1Image ReconstructionDynamic Scene Reconstruction