Pretrained Image-Text Models are Secretly Video Captioners
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910835494879232 |
|---|---|
| author | Zhang, Chunhui Jian, Yiren Ouyang, Zhongyu Vosoughi, Soroush |
| author_facet | Zhang, Chunhui Jian, Yiren Ouyang, Zhongyu Vosoughi, Soroush |
| contents | Developing video captioning models is computationally expensive. The dynamic nature of video also complicates the design of multimodal models that can effectively caption these sequences. However, we find that by using minimal computational resources and without complex modifications to address video dynamics, an image-based model can be repurposed to outperform several specialised video captioning systems. Our adapted model demonstrates top tier performance on major benchmarks, ranking 2nd on MSRVTT and MSVD, and 3rd on VATEX. We transform it into a competitive video captioner by post training a typical image captioning model BLIP2 with only 6,000 video text pairs and simply concatenating frames (significantly fewer data than other methods), which use 2.5 to 144 million pairs. From a resource optimization perspective, this video captioning study focuses on three fundamental factors: optimizing model scale, maximizing data efficiency, and incorporating reinforcement learning. This extensive study demonstrates that a lightweight, image based adaptation strategy can rival state-of-the-art video captioning systems, offering a practical solution for low-resource scenarios. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_13363 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Pretrained Image-Text Models are Secretly Video Captioners Zhang, Chunhui Jian, Yiren Ouyang, Zhongyu Vosoughi, Soroush Computer Vision and Pattern Recognition Machine Learning Developing video captioning models is computationally expensive. The dynamic nature of video also complicates the design of multimodal models that can effectively caption these sequences. However, we find that by using minimal computational resources and without complex modifications to address video dynamics, an image-based model can be repurposed to outperform several specialised video captioning systems. Our adapted model demonstrates top tier performance on major benchmarks, ranking 2nd on MSRVTT and MSVD, and 3rd on VATEX. We transform it into a competitive video captioner by post training a typical image captioning model BLIP2 with only 6,000 video text pairs and simply concatenating frames (significantly fewer data than other methods), which use 2.5 to 144 million pairs. From a resource optimization perspective, this video captioning study focuses on three fundamental factors: optimizing model scale, maximizing data efficiency, and incorporating reinforcement learning. This extensive study demonstrates that a lightweight, image based adaptation strategy can rival state-of-the-art video captioning systems, offering a practical solution for low-resource scenarios. |
| title | Pretrained Image-Text Models are Secretly Video Captioners |
| topic | Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2502.13363 |