6405 papers
This paper introduces SABER, a novel framework that generates spatially consistent, universal 3D adversarial objects to realistically and effectively attack Bird's-Eye-View detectors in autonomous driving by optimizing non-invasive environmental manipulations that maintain multi-view and temporal consistency.
This paper introduces Interaction Field Matching (IFM), a generalized framework that overcomes the modeling complexities of Electrostatic Field Matching by leveraging a novel interaction field inspired by strong quark-antiquark interactions to improve data generation and transfer performance.
This paper introduces HSSBench, a comprehensive multilingual benchmark featuring over 13,000 samples generated through a novel expert-agent collaboration pipeline, designed to evaluate and address the current limitations of Multimodal Large Language Models in handling the interdisciplinary and abstract reasoning tasks characteristic of the Humanities and Social Sciences.
This paper introduces Frame Guidance, a training-free method that enables fine-grained, frame-level control over video generation in diffusion models through efficient latent processing and optimization, eliminating the need for costly fine-tuning while supporting diverse tasks like keyframe guidance, stylization, and looping.
Perception-R1 addresses the limitation of existing RLVR methods in enhancing multimodal perception by introducing a novel visual perception reward derived from Chain-of-Thought annotations, which effectively boosts both perception and reasoning capabilities of Multimodal Large Language Models to achieve state-of-the-art performance with minimal training data.
StreamSplat is a fully feed-forward framework that enables real-time, online reconstruction of dynamic 3D scenes from uncalibrated video streams into 3D Gaussian Splatting representations, achieving state-of-the-art quality with a 1200x speedup over traditional optimization-based methods through probabilistic sampling, bidirectional deformation, and adaptive Gaussian fusion.
The paper proposes ECAD, a genetic algorithm-based evolutionary caching method that learns optimal, model-specific inference schedules to significantly accelerate off-the-shelf diffusion models while maintaining high image quality and generalizing across resolutions and architectures without requiring parameter modifications.
This paper demonstrates that generating images on-the-fly via Text-to-Image models can unlock latent visual priors to significantly enhance text-centric reasoning, provided there is strong semantic alignment, task visual groundability, and high generative fidelity.
SceneStreamer is a unified autoregressive transformer framework that generates continuous, long-horizon traffic scenarios by predicting sequences of tokens representing dynamic elements like agents and traffic signals, thereby enabling the creation of realistic, diverse, and adaptive environments that significantly improve the robustness and generalization of autonomous driving policies.
This paper proposes a novel, memory-efficient annealing guidance scheduler that dynamically adjusts the guidance scale during diffusion sampling based on conditional noisy signals, thereby significantly improving both image quality and text alignment without requiring additional activations.
This paper proposes MC-INR, a novel framework that leverages meta-learning, dynamic error-based re-clustering, and a branched architecture to efficiently encode complex multivariate scientific simulation data on unstructured grids, overcoming the inflexibility and single-variable limitations of existing Implicit Neural Representation methods.
CoBELa is a decoder-free, energy-based framework that enables transparent and compositional generative control by conditioning a frozen pretrained generator entirely through additive per-concept energy functions, achieving high image quality and concept accuracy without requiring model retraining.
InstructVLA introduces a novel Vision-Language-Action Instruction Tuning paradigm that successfully bridges flexible multimodal reasoning and precise manipulation by jointly optimizing embodied reasoning and action generation, thereby achieving state-of-the-art performance in both simulated and real-world robotic tasks without sacrificing pre-trained capabilities.
DMTrack introduces a novel dual-adapter architecture featuring a spatio-temporal modality adapter and a progressive modality complementary adapter to achieve state-of-the-art spatio-temporal multimodal tracking performance with only 0.93M trainable parameters.
This paper proposes a novel zero-shot 3D CT super-resolution framework that combines diffusion-based 2D projection priors with a signed 3D Gaussian splatting method featuring Negative Alpha Blending to reconstruct high-resolution volumes from single low-resolution inputs without requiring paired training data.
MMTok is a novel method that enhances the inference efficiency of Vision-Language Models by formulating vision token selection as a maximum coverage problem, leveraging complementary multimodal information from both vision and text to prune redundant tokens while preserving high performance.
This paper introduces ConEQsA, a novel agentic framework and benchmark designed to address the challenges of concurrent and asynchronous embodied question answering by leveraging shared memory and urgency-aware scheduling to outperform traditional sequential approaches in realistic multi-question scenarios.
This paper systematically evaluates Vision-Language Models' capabilities in autonomous driving lane topology awareness through a new BEV-based diagnostic framework, revealing that while performance correlates with model size and reasoning depth, current models—including frontier closed-source systems—still struggle with fundamental spatial reasoning tasks essential for safe navigation.
The paper introduces SiNGER, a novel knowledge distillation framework that utilizes singular nullspace-guided energy reallocation via a LoRA-based adapter to suppress high-norm artifacts in Vision Transformer teachers while preserving informative signals, thereby enabling student models to achieve state-of-the-art performance and clearer representations.
The paper introduces Earth-Agent, a novel agentic framework that unifies RGB and spectral Earth observation data within an MCP-based tool ecosystem to enable complex, multi-step quantitative reasoning, accompanied by the Earth-Bench benchmark for comprehensive evaluation of such capabilities.