Research map / Video / Video generation
STRIVE: Scene Text Replacement in Videos
Vijay Kumar B G, Jeyasri Subramanian, Varnith Chordia, Eugene Bart, Shaobo Fang, Kelly Guan, Raja Bala
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 reference frame selection instead of using the first frame (ablation) | STRIVE · Synthtext | MSE: 0.0203 -> 0.0186 | Improved |
| add reference frame selection instead of using the first frame (ablation) | STRIVE · Synthtext | PSNR: 16.93 -> 17.31 | Improved |
| add reference frame selection instead of using the first frame (ablation) | STRIVE · Synthtext | SSIM: 0.596 -> 0.604 | Improved |
| add blur prediction network (BPN) (ablation) | STRIVE · Synthtext | MSE: 0.0203 -> 0.0186 | Improved |
| add blur prediction network (BPN) (ablation) | STRIVE · Synthtext | SSIM: 0.594 -> 0.604 | Improved |
| replace per-frame SRNet with reference-frame replacement plus pose normalization and text propagation | SRNet · Synthtext | PSNR: 16.44 -> 17.31 | Improved |
| replace per-frame SRNet with STRIVE | SRNet · Robotext and Realworld | OCR accuracy: 0.771 -> 0.957 | Improved |
| replace per-frame SRNet with STRIVE (temporal consistency constraints) | SRNet · Robotext and Realworld | Jitter: 5.10 -> 1.48 | Improved |
| replace per-frame SRNet with STRIVE | SRNet · Inference speed, K=100 outputs | frames per second: 1.67 -> 8.01 | Improved |
| add +/- 5 pixel x-y misalignment to BPN input (ablation) | BPN · BPN simulation | MAE of w: 0.14 -> 0.15 | No clear effect |
| replace pix2pix with SRNet | pix2pix · Synthtext | PSNR: 12.27 -> 16.44 | Improved |
Closest papers by meaning
- DreamText: High Fidelity Scene Text Synthesis (CVPR 2025 · similarity 0.85)
- Inference-Time Text-to-Video Alignment with Diffusion Latent Beam Search (NeurIPS 2025 · similarity 0.85)
- Space-Time Diffusion Features for Zero-Shot Text-Driven Motion Transfer (CVPR 2024 · similarity 0.85)
- ViTeX-Bench: Benchmarking High-Fidelity Video Scene Text Editing (NeurIPS 2026 · similarity 0.85)
- SteerVTE: Seamless Video Text Editing with Style and Glyph Control (NeurIPS 2026 · similarity 0.84)
- ReplaceAnything3D: Text-Guided Object Replacement in 3D Scenes with Compositional Scene Representations (NeurIPS 2024 · similarity 0.84)
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.
