SynPO: Synergizing Descriptiveness and Preference Optimization for Video Detailed Captioning

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Dang, Jisheng, Zhang, Yizhou, Ye, Hao, Wang, Teng, Chen, Siming, Zheng, Huicheng, Guo, Yulan, Lai, Jianhuang, Hu, Bin
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915879213596672
author Dang, Jisheng
Zhang, Yizhou
Ye, Hao
Wang, Teng
Chen, Siming
Zheng, Huicheng
Guo, Yulan
Lai, Jianhuang
Hu, Bin
author_facet Dang, Jisheng
Zhang, Yizhou
Ye, Hao
Wang, Teng
Chen, Siming
Zheng, Huicheng
Guo, Yulan
Lai, Jianhuang
Hu, Bin
contents Fine-grained video captioning aims to generate detailed, temporally coherent descriptions of video content. However, existing methods struggle to capture subtle video dynamics and rich detailed information. In this paper, we leverage preference learning to enhance the performance of vision-language models in fine-grained video captioning, while mitigating several limitations inherent to direct preference optimization (DPO). First, we propose a pipeline for constructing preference pairs that leverages the intrinsic properties of VLMs along with partial assistance from large language models, achieving an optimal balance between cost and data quality. Second, we propose Synergistic Preference Optimization (SynPO), a novel optimization method offering significant advantages over DPO and its variants. SynPO prevents negative preferences from dominating the optimization, explicitly preserves the model's language capability to avoid deviation of the optimization objective, and improves training efficiency by eliminating the need for the reference model. We extensively evaluate SynPO not only on video captioning benchmarks (e.g., VDC, VDD, VATEX) but also across well-established NLP tasks, including general language understanding and preference evaluation, using diverse pretrained models. Results demonstrate that SynPO consistently outperforms DPO variants while achieving 20\% improvement in training efficiency. Code is available at https://github.com/longmalongma/SynPO
format Preprint
id arxiv_https___arxiv_org_abs_2506_00835
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SynPO: Synergizing Descriptiveness and Preference Optimization for Video Detailed Captioning
Dang, Jisheng
Zhang, Yizhou
Ye, Hao
Wang, Teng
Chen, Siming
Zheng, Huicheng
Guo, Yulan
Lai, Jianhuang
Hu, Bin
Artificial Intelligence
Computer Vision and Pattern Recognition
Fine-grained video captioning aims to generate detailed, temporally coherent descriptions of video content. However, existing methods struggle to capture subtle video dynamics and rich detailed information. In this paper, we leverage preference learning to enhance the performance of vision-language models in fine-grained video captioning, while mitigating several limitations inherent to direct preference optimization (DPO). First, we propose a pipeline for constructing preference pairs that leverages the intrinsic properties of VLMs along with partial assistance from large language models, achieving an optimal balance between cost and data quality. Second, we propose Synergistic Preference Optimization (SynPO), a novel optimization method offering significant advantages over DPO and its variants. SynPO prevents negative preferences from dominating the optimization, explicitly preserves the model's language capability to avoid deviation of the optimization objective, and improves training efficiency by eliminating the need for the reference model. We extensively evaluate SynPO not only on video captioning benchmarks (e.g., VDC, VDD, VATEX) but also across well-established NLP tasks, including general language understanding and preference evaluation, using diverse pretrained models. Results demonstrate that SynPO consistently outperforms DPO variants while achieving 20\% improvement in training efficiency. Code is available at https://github.com/longmalongma/SynPO
title SynPO: Synergizing Descriptiveness and Preference Optimization for Video Detailed Captioning
topic Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2506.00835