Spatial understanding is a critical capability for LVLMs (Large Vision-Language Models) to advance embodied AI applications. Existing works primarily focus on enhancing spatial understanding within a single frame, i.e., injecting 3D spatial concepts into LVLMs under single coordinate system. However, such improvements struggle in real-world tasks that require consistent cross-view spatial reasoning. In this paper, we propose CVVG-Reasoner(Cross-View Visual Geometries) that lifts single-frame spatial comprehension to unified cross-view spatial understanding by mimicking human-like cross-view reasoning mechanisms. First, we introduce MV3DSR(Multi-View 3D Spatial Reasoning), a scalable pipeline for cross-view spatial reasoning data generation, and construct MV3DSR-Dataset, a large-scale dataset with diverse 3D cross-view reasoning tasks. Based on MV3DSR, we propose MV3DSR-Bench, a comprehensive benchmark for evaluating cross-view spatial reasoning capabilities. Second, we design a three-stage training strategy: the first two stages progressively equip the model with (1) fundamental spatial knowledge and (2) human-like cross-view reasoning patterns, while the final stage employs reinforcement learning to further boost its performance. Extensive experiments demonstrate that our CVVG-Reasoner significantly outperforms existing 3D LLMs(Large Language Models) and advanced LVLMs in cross-view tasks while maintaining robust performance on out-of-domain data. Ablations further reveal that injecting human-like reasoning patterns yields 44% performance gain, validating the effectiveness of our design.

Language-Visual Large Models (LVLMs) have made significant strides in enhancing visual understanding capabilities. However, these models often struggle with knowledge-based visual tasks due to constrains in their pre-training data scope and timeliness. Existing Retrieval-Augmented Generation (RAG) methods can effectively solve the problem but primarily rely on user queries, limiting their applicability in scenarios without explicit language input. To overcome these challenges, we introduce MsRAG, a knowledge-augmented captioning framework designed to effectively retrieve and utilize external real-world knowledge, particularly in the absence of user queries, and perform dense captioning for subjects. MsRAG comprises three key components: (1) Parallel Visual Search Module. It retrieves fine-grained object-level knowledge using both online visual search engines and offline domain-knowledge databases, enhancing the robustness and richness of retrieved information. (2) Prompt Templates Pool. The prompt pool dynamically assigns appropriate prompts based on retrieved information, optimizing LVLMs' ability to leverage relevant data under complex RAG conditions. (3) Visual-RAG Alignment Module, which employs a novel visual prompting method to bridge the modality gap between textual RAG content and corresponding visual objects, enabling precise alignment of visual elements with their text-format RAG content. To validate the effectiveness of MsRAG, we conducted a series of qualitative and quantitative experiments. The evaluation results demonstrate the superiority of MsRAG over other methods.

Image editing approaches with diffusion models have been rapidly developed, yet their applicability are subject to requirements such as specific editing types (e.g., foreground or background object editing, style transfer), multiple conditions (e.g., mask, sketch, caption), and time consuming fine-tuning of diffusion models. For alleviating these limitations and realizing efficient real image editing, we propose a novel editing technique that only requires an input image and target text for various editing types including non-rigid edits without fine-tuning diffusion model. Our method contains three novelties: (I) Target-text Inversion Schedule (TTIS) is designed to fine-tune the input target text embedding to achieve fast image reconstruction without image caption and acceleration of convergence. (II) Progressive Transition Scheme applies progressive linear interpolation between target text embedding and its fine-tuned version to generate transition embedding for maintaining non-rigid editing capability. (III) Balanced Attention Module (BAM) balances the tradeoff between textual description and image semantics. By the means of combining self-attention map from reconstruction process and cross-attention map from transition process, the guidance of target text embeddings in diffusion process is optimized. In order to demonstrate editing capability, effectiveness and efficiency of the proposed BARET, we have conducted extensive qualitative and quantitative experiments. Moreover, results derived from user study and ablation study further prove the superiority over other methods.