Research map / Video / Video segmentation and tracking
Learning Omnidirectional Flow in 360° Video via Siamese Representation
Keshav Bhandari, Bin Duan, Gaowen Liu, Hugo Latapie, Ziliang Zong, Yan Yan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add SLOF siamese rotation-augmented training (v1) to RAFT, versus fine-tuning RAFT | RAFT · FLOW360 test (all speeds) | EPE: 1.624 -> 1.568 | Improved |
| add SLOF siamese training (v1) versus fine-tuned RAFT | RAFT · FLOW360 test (distortion-weighted, all speeds) | EPE_d: 2.635 -> 2.548 | Improved |
| add SLOF siamese training (v2) versus fine-tuned RAFT | RAFT · FLOW360 test (distortion-weighted, all speeds) | AE_d: 0.745 -> 0.691 | Improved |
| use SLOF (v1) instead of kernel transformer convolution adaptation | RAFT · FLOW360 test (distortion-weighted, all speeds) | EPE_d: 3.899 -> 2.548 | Improved |
| use KTN spherical convolution adaptation on pretrained RAFT | RAFT · FLOW360 test (all speeds) | EPE: 2.058 -> 2.222 | Hurt |
| fine-tune RAFT on FLOW360 | RAFT · FLOW360 test (all speeds) | EPE: 2.058 -> 1.624 | Improved |
| switch rotation between streams (v2) instead of single fixed rotation (v1) (ablation) | SLOF · FLOW360 test (all speeds) | EPE: 1.568 -> 1.615 | No clear effect |
| switch rotation between streams (v2) instead of single fixed rotation (v1) (ablation) | SLOF · FLOW360 test (all speeds) | AE: 0.497 -> 0.485 | No clear effect |
| add rotational invariance training (ablation) | LiteFlowNet360 · FLOW360 (rotational invariance study) | EPE: 3.95 -> 2.52 | Improved |
| add rotational invariance training (ablation) | Tangent Images · FLOW360 (rotational invariance study) | EPE: 3.57 -> 1.78 | Improved |
Closest papers by meaning
- Learning Optical Flow From Still Images (CVPR 2021 · similarity 0.89)
- Taming generative video models for zero-shot optical flow extraction (NeurIPS 2025 · similarity 0.88)
- DPFlow: Adaptive Optical Flow Estimation with a Dual-Pyramid Framework (CVPR 2025 · similarity 0.87)
- OCAI: Improving Optical Flow Estimation by Occlusion and Consistency Aware Interpolation (CVPR 2024 · similarity 0.86)
- HuPerFlow: A Comprehensive Benchmark for Human vs. Machine Motion Estimation Comparison (CVPR 2025 · similarity 0.86)
- UnSAMFlow: Unsupervised Optical Flow Guided by Segment Anything Model (CVPR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
