Research map / Detection and segmentation / CNN features
Zoom-In-To-Check: Boosting Video Interpolation via Instance-Level Discrimination
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 instance-level adversarial discrimination instead of whole-image adversarial discrimination (ablation) | Ours video interpolation model · CityScapes | SSIM: 0.902 -> 0.925 | Helped |
| add instance-level adversarial discrimination instead of whole-image adversarial discrimination (ablation) | Ours video interpolation model · UCF101 | SSIM: 0.870 -> 0.882 | Helped |
| add instance-level adversarial discrimination (ablation) | Ours video interpolation model · CityScapes | IE: 9.38 -> 8.03 | Helped |
| add instance-level adversarial discrimination (ablation) | Ours video interpolation model · CityScapes | PSNR: 29.31 -> 30.77 | Helped |
| add instance-level adversarial discrimination (ablation) | Ours video interpolation model · UCF101 | IE: 11.23 -> 10.92 | Helped |
| add instance-level adversarial discrimination (ablation) | Ours video interpolation model · UCF101 | PSNR: 30.08 -> 30.23 | Helped |
| add whole-image adversarial loss without instance-focused crops (ablation) | Ours video interpolation model · UCF101 | SSIM: 0.876 -> 0.870 | Hurt |
| add whole-image adversarial loss without instance-focused crops (ablation) | Ours video interpolation model · CityScapes | SSIM: 0.890 -> 0.902 | Helped |
| train the instance discriminator using high-resolution real patches corresponding to low-resolution RoIs (ablation) | Ours instance-level adversarial model · CityScapes object instances | interpolation quality | Helped |
| use instance-level discrimination rather than whole-image discrimination (ablation) | Ours video interpolation model · CityScapes trimaps wider than 12 pixels | IE | Hurt |
| use instance-level discrimination rather than whole-image discrimination (ablation) | Ours video interpolation model · CityScapes trimaps wider than 12 pixels | PSNR | Hurt |
| vary the number of region proposals per training image from 10 to 30 (ablation) | Ours instance-level adversarial model · test datasets | SSIM: ±0.002 | No clear effect |
Closest papers by meaning
- Depth-Aware Video Frame Interpolation (CVPR 2019 · similarity 0.90)
- Super SloMo: High Quality Estimation of Multiple Intermediate Frames for Video Interpolation (CVPR 2018 · similarity 0.87)
- Video Frame Interpolation Transformer (CVPR 2022 · similarity 0.87)
- Context-Aware Synthesis for Video Frame Interpolation (CVPR 2018 · similarity 0.87)
- Supervision Interpolation via LossMix: Generalizing Mixup for Object Detection and Beyond (AAAI 2024 · similarity 0.87)
- OCAI: Improving Optical Flow Estimation by Occlusion and Consistency Aware Interpolation (CVPR 2024 · 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.