Scene Graph-guided SegCaptioning Transformer with Fine-grained Alignment for Controllable Video Segmentation and Captioning

2026-03-21

AI Summary

This work addresses the limitation of existing video multimodal understanding methods in responding to users’ fine-grained intent regarding specific spatial regions. The authors propose a controllable video segmentation and captioning task that, given user-provided prompts such as bounding boxes, jointly generates precise segmentation masks and corresponding textual descriptions. The core innovation lies in a scene graph–guided fine-grained alignment mechanism that enables token-level co-prediction of masks and language. To model user intent effectively, they introduce a Prompt-guided Temporal Graph Former, coupled with a Fine-grained Mask-linguistic Decoder and a multi-entity contrastive loss for joint optimization. Experiments demonstrate that the proposed method significantly outperforms state-of-the-art approaches on two benchmark datasets, accurately capturing user intent and producing high-quality, customizable multimodal outputs.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Recent advancements in multimodal large models have significantly bridged the representation gap between diverse modalities, catalyzing the evolution of video multimodal interpretation, which enhances users' understanding of video content by generating correlated modalities. However, most existing video multimodal interpretation methods primarily concentrate on global comprehension with limited user interaction. To address this, we propose a novel task, Controllable Video Segmentation and Captioning (SegCaptioning), which empowers users to provide specific prompts, such as a bounding box around an object of interest, to simultaneously generate correlated masks and captions that precisely embody user intent. An innovative framework Scene Graph-guided Fine-grained SegCaptioning Transformer (SG-FSCFormer) is designed that integrates a Prompt-guided Temporal Graph Former to effectively captures and represents user intent through an adaptive prompt adaptor, ensuring that the generated content well aligns with the user's requirements. Furthermore, our model introduces a Fine-grained Mask-linguistic Decoder to collaboratively predict high-quality caption-mask pairs using a Multi-entity Contrastive loss, as well as provide fine-grained alignment between each mask and its corresponding caption tokens, thereby enhancing users' comprehension of videos. Comprehensive experiments conducted on two benchmark datasets demonstrate that SG-FSCFormer achieves remarkable performance, effectively capturing user intent and generating precise multimodal outputs tailored to user specifications. Our code is available at https://github.com/XuZhang1211/SG-FSCFormer.

Problem

• Controllable Video Segmentation
• Video Captioning
• User Intent
• Multimodal Interpretation
• Fine-grained Alignment

Innovation

• Controllable Video Segmentation
• Scene Graph
• Fine-grained Alignment
• Prompt-guided Generation
• Multimodal Transformer