Research map / Video / Video segmentation and tracking
CDFI: Compression-Driven Network Design for Frame 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 |
|---|---|---|---|
| compress AdaCoF using sparsity-induced pruning (ablation) | AdaCoF · Middlebury | PSNR: 35.72 -> 35.43 | No clear effect |
| compress AdaCoF using sparsity-induced pruning (ablation) | AdaCoF · model size | Parameters (M): 21.8 -> 2.45 | Helped |
| add a feature pyramid and synthesis network without path selection (ablation) | Compressed AdaCoF · Middlebury | PSNR: 35.43 -> 36.13 | Helped |
| add a feature pyramid and synthesis network without path selection (ablation) | Compressed AdaCoF · Vimeo-90K | LPIPS: 0.020 -> 0.011 | Helped |
| add 1x1 convolutions to filter multiscale features (ablation) | CDFI: FP · Middlebury | PSNR: 36.13 -> 36.52 | Helped |
| increase the AdaCoF kernel setting from F=5, d=1 to F=11, d=2 (ablation) | CDFI: FP + 1x1 Conv · Middlebury | PSNR: 36.52 -> 36.70 | Helped |
| add path selection to combine the AdaCoF output and synthesis-network output (ablation) | CDFI: FP + 1x1 Conv · Middlebury | PSNR: 36.70 -> 37.14 | Helped |
| add path selection to combine the AdaCoF output and synthesis-network output (ablation) | CDFI: FP + 1x1 Conv · UCF101-DVF | PSNR: 35.14 -> 35.21 | Helped |
| add path selection to combine the AdaCoF output and synthesis-network output (ablation) | CDFI: FP + 1x1 Conv · Vimeo-90K | PSNR: 34.82 -> 35.17 | Helped |
| use CDFI rather than AdaCoF | AdaCoF · Middlebury | PSNR: 35.72 -> 37.14 | Helped |
| use CDFI rather than AdaCoF | AdaCoF · Vimeo-90K | PSNR: 34.35 -> 35.17 | Helped |
| use CDFI rather than AdaCoF | AdaCoF · UCF101-DVF | PSNR: 35.16 -> 35.21 | Helped |
Closest papers by meaning
- ST-MFNet: A Spatio-Temporal Multi-Flow Network for Frame Interpolation (CVPR 2022 · similarity 0.87)
- A Unified Pyramid Recurrent Network for Video Frame Interpolation (CVPR 2023 · similarity 0.86)
- VFIMamba: Video Frame Interpolation with State Space Models (NeurIPS 2024 · similarity 0.86)
- LDMVFI: Video Frame Interpolation with Latent Diffusion Models (AAAI 2024 · similarity 0.85)
- High-Resolution Frame Interpolation with Patch-based Cascaded Diffusion (AAAI 2025 · similarity 0.85)
- Repurposing Pre-trained Video Diffusion Models for Event-based Video Interpolation (CVPR 2025 · similarity 0.85)
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.