LOVE: Benchmarking and Evaluating Text-to-Video Generation and Video-to-Text Interpretation

Jiarui Wang, Huiyu Duan, Ziheng Jia, Zicheng Zhang, Yu Zhao, Juntong Wang, Guangtao Zhai, Xiongkuo Min
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:126046-126057, 2026.

Abstract

Recent advancements in large multimodal models (LMMs) have driven substantial progress in both text-to-video (T2V) generation and video-to-text (V2T) interpretation tasks. However, current AI-generated videos (AIGVs) still exhibit limitations in terms of perceptual quality and text-video alignment. To this end, we present AIGVE-60K, a comprehensive dataset and benchmark for AI-Generated Video Evaluation, which features (i) comprehensive tasks, encompassing 3,050 extensive prompts across 20 fine-grained task dimensions, (ii) the largest human annotations, including 120K mean-opinion scores (MOSs) and 60K question-answering (QA) pairs annotated on 58,500 videos generated from 30 T2V models, and (iii) bidirectional benchmarking and evaluating for both T2V generation and V2T interpretation capabilities. Based on AIGVE-60K, we propose LOVE, a LMM-based metric for AIGV Evaluation from multiple dimensions including perceptual preference, text-video correspondence, and task-specific accuracy. Building upon LOVE, we further introduce LOVE-Reward to optimize T2V models through reinforcement learning, effectively enhancing both the perceptual quality and text-video correspondence of generated videos. Comprehensive experiments demonstrate that LOVE achieves state-of-the-art performance and generalizes effectively to various AIGV benchmarks. LOVE-Reward significantly improves video generation quality. These findings highlight the effectiveness of the AIGVE-60K dataset and our proposed methods. The database and codes are available at https://github.com/IntMeGroup/LOVE.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-wang26aw, title = {{LOVE}: Benchmarking and Evaluating Text-to-Video Generation and Video-to-Text Interpretation}, author = {Wang, Jiarui and Duan, Huiyu and Jia, Ziheng and Zhang, Zicheng and Zhao, Yu and Wang, Juntong and Zhai, Guangtao and Min, Xiongkuo}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {126046--126057}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/wang26aw/wang26aw.pdf}, url = {https://proceedings.mlr.press/v306/wang26aw.html}, abstract = {Recent advancements in large multimodal models (LMMs) have driven substantial progress in both text-to-video (T2V) generation and video-to-text (V2T) interpretation tasks. However, current AI-generated videos (AIGVs) still exhibit limitations in terms of perceptual quality and text-video alignment. To this end, we present AIGVE-60K, a comprehensive dataset and benchmark for AI-Generated Video Evaluation, which features (i) comprehensive tasks, encompassing 3,050 extensive prompts across 20 fine-grained task dimensions, (ii) the largest human annotations, including 120K mean-opinion scores (MOSs) and 60K question-answering (QA) pairs annotated on 58,500 videos generated from 30 T2V models, and (iii) bidirectional benchmarking and evaluating for both T2V generation and V2T interpretation capabilities. Based on AIGVE-60K, we propose LOVE, a LMM-based metric for AIGV Evaluation from multiple dimensions including perceptual preference, text-video correspondence, and task-specific accuracy. Building upon LOVE, we further introduce LOVE-Reward to optimize T2V models through reinforcement learning, effectively enhancing both the perceptual quality and text-video correspondence of generated videos. Comprehensive experiments demonstrate that LOVE achieves state-of-the-art performance and generalizes effectively to various AIGV benchmarks. LOVE-Reward significantly improves video generation quality. These findings highlight the effectiveness of the AIGVE-60K dataset and our proposed methods. The database and codes are available at https://github.com/IntMeGroup/LOVE.} }
Endnote
%0 Conference Paper %T LOVE: Benchmarking and Evaluating Text-to-Video Generation and Video-to-Text Interpretation %A Jiarui Wang %A Huiyu Duan %A Ziheng Jia %A Zicheng Zhang %A Yu Zhao %A Juntong Wang %A Guangtao Zhai %A Xiongkuo Min %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-wang26aw %I PMLR %P 126046--126057 %U https://proceedings.mlr.press/v306/wang26aw.html %V 306 %X Recent advancements in large multimodal models (LMMs) have driven substantial progress in both text-to-video (T2V) generation and video-to-text (V2T) interpretation tasks. However, current AI-generated videos (AIGVs) still exhibit limitations in terms of perceptual quality and text-video alignment. To this end, we present AIGVE-60K, a comprehensive dataset and benchmark for AI-Generated Video Evaluation, which features (i) comprehensive tasks, encompassing 3,050 extensive prompts across 20 fine-grained task dimensions, (ii) the largest human annotations, including 120K mean-opinion scores (MOSs) and 60K question-answering (QA) pairs annotated on 58,500 videos generated from 30 T2V models, and (iii) bidirectional benchmarking and evaluating for both T2V generation and V2T interpretation capabilities. Based on AIGVE-60K, we propose LOVE, a LMM-based metric for AIGV Evaluation from multiple dimensions including perceptual preference, text-video correspondence, and task-specific accuracy. Building upon LOVE, we further introduce LOVE-Reward to optimize T2V models through reinforcement learning, effectively enhancing both the perceptual quality and text-video correspondence of generated videos. Comprehensive experiments demonstrate that LOVE achieves state-of-the-art performance and generalizes effectively to various AIGV benchmarks. LOVE-Reward significantly improves video generation quality. These findings highlight the effectiveness of the AIGVE-60K dataset and our proposed methods. The database and codes are available at https://github.com/IntMeGroup/LOVE.
APA
Wang, J., Duan, H., Jia, Z., Zhang, Z., Zhao, Y., Wang, J., Zhai, G. & Min, X.. (2026). LOVE: Benchmarking and Evaluating Text-to-Video Generation and Video-to-Text Interpretation. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:126046-126057 Available from https://proceedings.mlr.press/v306/wang26aw.html.

Related Material