Papers › Diverse Video Captioning by Adaptive Spatio-temporal Attention
Diverse Video Captioning by Adaptive Spatio-temporal Attention
Zohreh Ghaderi, Leonard Salewski, Hendrik P. A. Lensch
To generate proper captions for videos, the inference needs to identify relevant concepts and pay attention to the spatial relationships between them as well as to the temporal development in the clip. Our end-to-end encoder-decoder video captioning framework incorporates two transformer-based architectures, an adapted transformer for a single joint spatio-temporal video analysis as well as a self-attention-based decoder for advanced text generation. Furthermore, we introduce an adaptive frame selection scheme to reduce the number of required incoming frames while maintaining the relevant content when training both transformers. Additionally, we estimate semantic concepts relevant for video captioning by aggregating all ground truth captions of each sample. Our approach achieves state-of-the-art results on the MSVD, as well as on the large-scale MSR-VTT and the VATEX benchmark datasets considering multiple Natural Language Generation (NLG) metrics. Additional evaluations on diversity scores highlight the expressiveness and diversity in the structure of our generated captions.
Code
Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.
Code Syntology ran Syntology
Not run by Syntology. Nothing on this page verifies that the listed code works.
Tasks
Results from the paper archive 2025-07-28
| Task | Dataset | Model | Metric | Value | Rank at snapshot | Leaderboard | Report |
|---|---|---|---|---|---|---|---|
| Video Captioning | MSR-VTT | VASTA (Vatex-backbone) | BLEU-4 | 44.21 | #20 of 24 | Archive leaderboard | report |
| Video Captioning | MSR-VTT | VASTA (Vatex-backbone) | CIDEr | 56.08 | #20 of 24 | Archive leaderboard | report |
| Video Captioning | MSR-VTT | VASTA (Vatex-backbone) | METEOR | 30.24 | #20 of 24 | Archive leaderboard | report |
| Video Captioning | MSR-VTT | VASTA (Vatex-backbone) | ROUGE-L | 62.9 | #20 of 24 | Archive leaderboard | report |
| Video Captioning | MSR-VTT | VASTA (Kinetics-backbone) | BLEU-4 | 43.4 | #21 of 24 | Archive leaderboard | report |
| Video Captioning | MSR-VTT | VASTA (Kinetics-backbone) | CIDEr | 55 | #21 of 24 | Archive leaderboard | report |
| Video Captioning | MSR-VTT | VASTA (Kinetics-backbone) | METEOR | 30.2 | #21 of 24 | Archive leaderboard | report |
| Video Captioning | MSR-VTT | VASTA (Kinetics-backbone) | ROUGE-L | 62.5 | #21 of 24 | Archive leaderboard | report |
| Video Captioning | MSVD | VASTA (Vatex-backbone) | BLEU-4 | 59.2 | #12 of 16 | Archive leaderboard | report |
| Video Captioning | MSVD | VASTA (Vatex-backbone) | CIDEr | 119.7 | #12 of 16 | Archive leaderboard | report |
| Video Captioning | MSVD | VASTA (Vatex-backbone) | METEOR | 40.65 | #12 of 16 | Archive leaderboard | report |
| Video Captioning | MSVD | VASTA (Vatex-backbone) | ROUGE-L | 76.7 | #12 of 16 | Archive leaderboard | report |
| Video Captioning | MSVD | VASTA (Kinetics-backbone) | BLEU-4 | 56.1 | #15 of 16 | Archive leaderboard | report |
| Video Captioning | MSVD | VASTA (Kinetics-backbone) | CIDEr | 106.4 | #15 of 16 | Archive leaderboard | report |
| Video Captioning | MSVD | VASTA (Kinetics-backbone) | METEOR | 39.1 | #15 of 16 | Archive leaderboard | report |
| Video Captioning | MSVD | VASTA (Kinetics-backbone) | ROUGE-L | 74.5 | #15 of 16 | Archive leaderboard | report |
| Video Captioning | VATEX | VASTA (Kinetics-backbone) | BLEU-4 | 36.25 | #7 of 10 | Archive leaderboard | report |
| Video Captioning | VATEX | VASTA (Kinetics-backbone) | CIDEr | 65.07 | #7 of 10 | Archive leaderboard | report |
| Video Captioning | VATEX | VASTA (Kinetics-backbone) | METEOR | 25.32 | #7 of 10 | Archive leaderboard | report |
| Video Captioning | VATEX | VASTA (Kinetics-backbone) | ROUGE-L | 51.88 | #7 of 10 | Archive leaderboard | report |
Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections