Papers by Noel E. O’Connor
BERTHA: Video Captioning Evaluation Via Transfer-Learned Human Assessment (2022.lrec-1)
Copied to clipboard
| Challenge: | Existing metrics to evaluate video captioning systems are based on overlap between the caption and the reference sentence, but fail to include the context of the scene. |
| Approach: | They propose a method to evaluate video captioning systems using a deep learning model . the model is based on a language model that has been shown to work well in NLP tasks . |
| Outcome: | The proposed model outperforms the most commonly used metrics in video to text tasks. |