Research map / Video / Video-language understanding
Temporal Deformable Convolutional Encoder-Decoder Networks for Video Captioning
Read the paperCheck 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 |
|---|---|---|---|
| Replace the recurrent encoder-decoder with TDConvED | Video captioning with ResNet features · MSVD | CIDEr-D: 72.0 -> 76.4 | Helped |
| Replace the recurrent encoder-decoder with TDConvED | Video captioning with ResNet features · MSR-VTT | CIDEr-D: 42.1 -> 42.8 | Helped |
| Add a convolutional decoder to the MP-LSTM-style system (ablation) | TDConvED₁ · MSVD (V+C) | CIDEr-D: 55.1 -> 61.6 | Helped |
| Add a convolutional decoder to the MP-LSTM-style system (ablation) | TDConvED₁ · MSR-VTT (R) | CIDEr-D: 35.8 -> 40.7 | Helped |
| Add a temporal deformable convolutional encoder (ablation) | TDConvED₁ to form TDConvED₂ · MSVD (V+C) | CIDEr-D: 61.6 -> 65.3 | Helped |
| Add a temporal deformable convolutional encoder (ablation) | TDConvED₁ to form TDConvED₂ · MSR-VTT (R) | CIDEr-D: 40.7 -> 41.9 | Helped |
| Add temporal attention over contextually encoded video features (ablation) | TDConvED₂ to form TDConvED · MSVD (V+C) | CIDEr-D: 65.3 -> 67.2 | Helped |
| Add temporal attention over contextually encoded video features (ablation) | TDConvED₂ to form TDConvED · MSR-VTT (R) | CIDEr-D: 41.9 -> 42.8 | Helped |
| Use 2 decoder convolutional blocks rather than 1 (ablation) | TDConvED₁ · MSVD | CIDEr-D: 57.9 -> 61.6 | Helped |
| Use 3 rather than 2 decoder convolutional blocks (ablation) | TDConvED₁ · MSVD | CIDEr-D: 61.6 -> 59.8 | Hurt |
| Use 2 temporal deformable encoder blocks rather than 1 (ablation) | TDConvED₂ (V+C) · MSVD | CIDEr-D: 64.3 -> 65.3 | Helped |
| Use 3 rather than 2 temporal deformable encoder blocks (ablation) | TDConvED₂ (V+C) · MSVD | CIDEr-D: 65.3 -> 62.8 | Hurt |
Closest papers by meaning
- Hierarchical Recurrent Neural Encoder for Video Representation With Application to Captioning (CVPR 2016 · similarity 0.90)
- Spatio-Temporal Dynamics and Semantic Attribute Enriched Visual Encoding for Video Captioning (CVPR 2019 · similarity 0.90)
- Hierarchical Boundary-Aware Neural Encoder for Video Captioning (CVPR 2017 · similarity 0.89)
- Convolutional Image Captioning (CVPR 2018 · similarity 0.89)
- Reconstruction Network for Video Captioning (CVPR 2018 · similarity 0.88)
- Fine-Grained Captioning of Long Videos through Scene Graph Consolidation (ICML 2025 · 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.