Research map / Architectures and efficiency / RNNs and normalization
InterpoNet, a Brain Inspired Neural Network for Optical Flow Dense Interpolation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 lateral dependency loss (ablation) | InterpoNet · Sintel final validation | EPE: 6.104 -> 5.833 | Helped |
| Add multi-layer loss (ablation) | InterpoNet · Sintel final validation | EPE: 6.104 -> 5.656 | Helped |
| Combine lateral dependency and multi-layer losses (ablation) | InterpoNet · Sintel final validation | EPE: 6.104 -> 5.470 | Helped |
| Add edges map to sparse flow and mask inputs (ablation) | InterpoNet · Sintel final validation | EPE: 6.240 -> 5.470 | Helped |
| Fine-tune on Sintel FlowFields matches (ablation) | InterpoNet pretrained on Flying Chairs · Sintel final validation, FlowFields | EPE: 5.802 -> 5.470 | Helped |
| Fine-tune on the evaluation matching algorithm rather than FlowFields (ablation) | InterpoNet pretrained on Flying Chairs · Sintel final validation, CPM-Flow | EPE: 6.165 -> 5.851 | Helped |
| Pretrain with DeepMatching instead of FlowFields (ablation) | InterpoNet · Sintel final validation, DeepMatching | EPE: 5.971 -> 5.934 | Hurt |
| Use bi-directional averaging at evaluation (ablation) | InterpoNet · Sintel validation, FlowFields | EPE: 5.363 -> 5.470 | Hurt |
| Use bi-directional averaging at evaluation (ablation) | InterpoNet · Sintel validation, DeepMatching | EPE: 6.017 -> 5.971 | Helped |
| Replace EpicFlow interpolation with InterpoNet | FlowFields matches · Sintel final benchmark | EPE: 5.810 -> 5.535 | Helped |
| Replace EpicFlow interpolation with InterpoNet | DiscreteFlow matches · KITTI 2012 | EPE All: 3.6 -> 2.4 | Helped |
| Replace EpicFlow interpolation with InterpoNet | CPM-Flow matches · KITTI 2012 | %Out-All: 13.70 -> 14.57 | Hurt |
Closest papers by meaning
- Learning Optical Flow From a Few Matches (CVPR 2021 · similarity 0.88)
- Neuroexplicit Diffusion Models for Inpainting of Optical Flow Fields (ICML 2024 · similarity 0.87)
- Optical Flow Estimation Using a Spatial Pyramid Network (CVPR 2017 · similarity 0.87)
- FlowNet 2.0: Evolution of Optical Flow Estimation With Deep Networks (CVPR 2017 · similarity 0.87)
- Flow-NeRF: Joint Learning of Geometry, Poses, and Dense Flow within Unified Neural Representations (CVPR 2025 · similarity 0.86)
- Recurrent Partial Kernel Network for Efficient Optical Flow Estimation (AAAI 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.