paperlist.ai

AI 顶会论文

AI 顶会论文库

检索 30+ 个顶会的论文，包括 NeurIPS、ICLR、ICML、CVPR、ACL、AAAI 等。输入关键词筛选，打开任意一篇查看摘要、作者和链接。

arXiv

Evidence Recomposition and Predictive Context Residualization for Visual Attribution in Multimodal Large Language Models

Jiawei Liang, Jianjie Huang, Xianghao Jiao, Siyuan Liang, Shiming Liu, Xiaochun Cao

Multimodal large language models (MLLMs) have achieved strong vision-language performance, yet their token-level visual evidence remains difficult to inspect. Recent logit-lens attribution methods project each visual-token hidden state into the vocabulary space to explain generated words, but this token-wise readout introduces a mismatch: visual tokens are context-mixed by the model, while the attribution score is decoded independently at each token location. This often produces fragmented attribution maps and can be further affected by autoregressive context signals from preceding text tokens. We propose ERCR, an attribution framework built from Evidence Recomposition (ER) and Predictive Context Residualization (PCR). ER aggregates target evidence across multiple views with different token-to-region assignments, reducing attribution fragmentation caused by a single readout grid. PCR estimates a preceding-token context map with RBO-based rank relevance and subtracts its fitted component from the ER map to suppress context-token interference. Experiments on LLaVA, Qwen2-VL, and InternVL families across COCO Caption, GranDf, and OpenPSG show that ERCR improves visual evidence for target tokens and mitigates preceding-token context interference under the existing evaluation protocol. On Qwen2-VL-2B, ERCR improves TAM F1-IoU from 39.10 to 44.45 on COCO Caption and from 30.83 to 37.20 on GranDf. Overall, ERCR provides a practical refinement for token-level visual evidence inspection.

arXiv

Reprojection-Guided 3D Gaussian Splatting Diffusion for Weakly Supervised Single-Image Normal Estimation

Yanxing Liang, Yinghui Wang, Wei Li, Tao Yan, Jiaxing Shen

We propose CLONE, a Continuous Latent Optimization framework for Normal Estimation via 3D Gaussian splatting. The core idea is to construct an image-geometry-image consistency strategy that unifies explicit geometric representation with differentiable rendering, thereby enabling weakly supervised learning without normal ground truth. Specifically, CLONE comprises four components. First, by introducing a differentiable light interaction model with a learnable modulation kernel, we perform a unified reparameterization of the 3DGS parameter space. Second, the conditional single-step deterministic refinement network integrates denoising architectures with differentiable reprojection constraints to refine the initial normals, thereby adaptively recovering the high-frequency details erased by the inherently smooth Gaussian primitives. Third, the cross-domain gating fusion mechanism adaptively combines the two complementary normal estimates, reconciling the geometrically consistent yet over-smooth 3DGS estimate with the detailed yet potentially geometry-inconsistent refinement. Finally, all components are jointly optimized under a unified photometric reprojection objective with geometric consistency regularizations in a fully differentiable pathway, achieving an end-to-end optimization closed loop without relying on external normal labels.

arXiv

FuseLIP: Multimodal Embeddings via Early Fusion of Discrete Tokens

Christian Schlarmann, Francesco Croce, Nicolas Flammarion, Matthias Hein

Contrastive language-image pre-training aligns features of text-image pairs in a common latent space via distinct encoders for each modality. While this approach achieves impressive performance in several zero-shot tasks, it cannot natively handle multimodal inputs, i.e., encoding image and text into a single feature vector. As a remedy, it is common practice to use additional modules to merge the features extracted by unimodal encoders. In this work, we present FuseLIP, a new architecture for multimodal embedding. Leveraging recent progress in discrete image tokenizers, we propose to use a single transformer model operating on a unified vocabulary of text and image tokens. This early fusion approach allows the different modalities to interact at each depth of encoding and obtain richer representations compared to common late fusion. We collect new datasets for multimodal pre-training and evaluation, designing challenging tasks for multimodal encoders. We show that FuseLIP outperforms late fusion approaches in several multimodal and unimodal embedding tasks.

arXiv

Two by Two: Learning Multi-Task Pairwise Objects Assembly for Generalizable Robot Manipulation

Yu Qi, Yuanchen Ju, Tianming Wei, Chi Chu, Lawson L. S. Wong, Huazhe Xu

3D assembly tasks, such as furniture assembly and component fitting, play a crucial role in daily life and represent essential capabilities for future home robots. Existing benchmarks and datasets predominantly focus on assembling geometric fragments or factory parts, which fall short in addressing the complexities of everyday object interactions and assemblies. To bridge this gap, we present 2BY2, a large-scale annotated dataset for daily pairwise objects assembly, covering 18 fine-grained tasks that reflect real-life scenarios, such as plugging into sockets, arranging flowers in vases, and inserting bread into toasters. 2BY2 dataset includes 1,034 instances and 517 pairwise objects with pose and symmetry annotations, requiring approaches that align geometric shapes while accounting for functional and spatial relationships between objects. Leveraging the 2BY2 dataset, we propose a two-step SE(3) pose estimation method with equivariant features for assembly constraints. Compared to previous shape assembly methods, our approach achieves state-of-the-art performance across all 18 tasks in the 2BY2 dataset. Additionally, robot experiments further validate the reliability and generalization ability of our method for complex 3D assembly tasks.

arXiv

Helios 2.0: A Robust, Ultra-Low Power Gesture Recognition System Optimised for Event-Sensor based Wearables

Prarthana Bhattacharyya, Joshua Mitton, Ryan Page, Owen Morgan, Oliver Powell, Benjamin Menzies, Gabriel Homewood, Kemi Jacobs, Paolo Baesso, Taru Muhonen, Richard Vigars, Louis Berridge

We present an advance in wearable technology: a mobile-optimized, real-time, ultra-low-power event camera system that enables natural hand gesture control for smart glasses, dramatically improving user experience. While hand gesture recognition in computer vision has advanced significantly, critical challenges remain in creating systems that are intuitive, adaptable across diverse users and environments, and energy-efficient enough for practical wearable applications. Our approach tackles these challenges through carefully selected microgestures: lateral thumb swipes across the index finger (in both directions) and a double pinch between thumb and index fingertips. These human-centered interactions leverage natural hand movements, ensuring intuitive usability without requiring users to learn complex command sequences. To overcome variability in users and environments, we developed a novel simulation methodology that enables comprehensive domain sampling without extensive real-world data collection. Our power-optimised architecture maintains exceptional performance, achieving F1 scores above 80\% on benchmark datasets featuring diverse users and environments. The resulting models operate at just 6-8 mW when exploiting the Qualcomm Snapdragon Hexagon DSP, with our 2-channel implementation exceeding 70\% F1 accuracy and our 6-channel model surpassing 80\% F1 accuracy across all gesture classes in user studies. These results were achieved using only synthetic training data. This improves on the state-of-the-art for F1 accuracy by 20\% with a power reduction 25x when using DSP. This advancement brings deploying ultra-low-power vision systems in wearable devices closer and opens new possibilities for seamless human-computer interaction.

arXiv

RedDiffuser: Auditing Multimodal Safety Failures in Vision-Language Models via Reinforced Diffusion

Ruofan Wang, Xingjun Ma

Large Vision-Language Models (VLMs) are increasingly deployed in open-ended environments, where ensuring reliable safety under multimodal inputs is critical. However, existing evaluations remain largely instruction-centric, focusing on explicit malicious queries while overlooking a more realistic and underexplored risk: whether safety alignment remains robust under harmful contextual exposure. This limitation is particularly important for multimodal systems, where visual inputs can substantially steer model behavior and render text-only auditing insufficient. In this work, we study multimodal safety auditing under harmful contextual exposure, asking whether VLMs can maintain safe behavior when partial toxic text is paired with visual context. To enable systematic auditing, we propose RedDiffuser (RedDiff), a reinforcement-based framework that leverages diffusion models to generate semantically coherent visual inputs for black-box safety testing. By combining greedy prompt search with reinforcement optimization, RedDiffuser uncovers high-risk multimodal inputs that expose latent safety failures. Extensive experiments on both open-source and commercial VLMs show that such context-conditioned failures are widespread. On LLaVA, RedDiffuser increases unsafe response rates by up to 10.69% on the original set and 8.91% on a hold-out set, with strong transferability to Gemini and LLaMA-Vision. These vulnerabilities persist even under external safety guardrails, suggesting that current system-level safety mechanisms remain insufficient for realistic multimodal risks. Our findings reveal a critical blind spot in existing safety evaluations and establish context-aware multimodal auditing as an essential paradigm for diagnosing hidden vulnerabilities in modern VLM systems.

arXiv

GestureLSM: Latent Shortcut based Co-Speech Gesture Generation with Spatial-Temporal Modeling

Pinxin Liu, Luchuan Song, Junhua Huang, Haiyang Liu, Junfan Zhu, Chenliang Xu

Generating full-body human gestures based on speech signals remains challenges on quality and speed. Existing approaches model different body regions such as body, legs and hands separately, which fail to capture the spatial interactions between them and result in unnatural and disjointed movements. Additionally, their autoregressive/diffusion-based pipelines show slow generation speed due to dozens of inference steps. To address these two challenges, we propose GestureLSM, a flow-matching-based approach for Co-Speech Gesture Generation with spatial-temporal modeling. Our method i) explicitly model the interaction of tokenized body regions through spatial and temporal attention, for generating coherent full-body gestures. ii) introduce the flow matching to enable more efficient sampling by explicitly modeling the latent velocity space. To overcome the suboptimal performance of flow matching baseline, we propose latent shortcut learning and beta distribution time stamp sampling during training to enhance gesture synthesis quality and accelerate inference. Combining the spatial-temporal modeling and improved flow matching-based framework, GestureLSM achieves state-of-the-art performance on BEAT2 while significantly reducing inference time compared to existing methods, highlighting its potential for enhancing digital humans and embodied agents in real-world applications. Project Page: https://andypinxinliu.github.io/GestureLSM

arXiv

Rethinking Token-wise Feature Caching: Accelerating Diffusion Transformers with Dual Feature Caching

Chang Zou, Shikang Zheng, Evelyn Zhang, Runlin Guo, Haohang Xu, Zhengyi Shi, Conghui He, Xuming Hu, Linfeng Zhang

Diffusion Transformers (DiT) have become the dominant methods in image and video generation yet still suffer substantial computational costs. As an effective approach for DiT acceleration, feature caching methods are designed to cache the features of DiT in previous timesteps and reuse them in the next timesteps, allowing us to skip the computation in the next timesteps. Among them, token-wise feature caching has been introduced to perform different caching ratios for different tokens in DiTs, aiming to skip the computation for unimportant tokens while still computing the important ones. In this paper, we propose to carefully check the effectiveness in token-wise feature caching with the following two questions: (1) Is it really necessary to compute the so-called "important" tokens in each step? (2) Are so-called important tokens really important? Surprisingly, this paper gives some counter-intuition answers, demonstrating that consistently computing the selected ``important tokens'' in all steps is not necessary. The selection of the so-called ``important tokens'' is often ineffective, and even sometimes shows inferior performance than random selection. Based on these observations, this paper introduces dual feature caching referred to as DuCa, which performs aggressive caching strategy and conservative caching strategy iteratively and selects the tokens for computing randomly. Extensive experimental results demonstrate the effectiveness of our method in DiT, PixArt, FLUX, and OpenSora, demonstrating significant improvements than the previous token-wise feature caching.

arXiv

MiniGPT-Reverse-Designing: Predicting Image Adjustments Utilizing MiniGPT-4

Vahid Azizi, Fatemeh Koochaki

Vision-Language Models (VLMs) have recently seen significant advancements through integrating with Large Language Models (LLMs). The VLMs, which process image and text modalities simultaneously, have demonstrated the ability to learn and understand the interaction between images and texts across various multi-modal tasks. Reverse designing, which could be defined as a complex vision-language task, aims to predict the edits and their parameters, given a source image, an edited version, and an optional high-level textual edit description. This task requires VLMs to comprehend the interplay between the source image, the edited version, and the optional textual context simultaneously, going beyond traditional vision-language tasks. In this paper, we extend and fine-tune MiniGPT-4 for the reverse designing task. Our experiments demonstrate the extensibility of off-the-shelf VLMs, specifically MiniGPT-4, for more complex tasks such as reverse designing. Code is available at this \href{https://github.com/VahidAz/MiniGPT-Reverse-Designing}{repository}.

arXiv

Partial FC: Training 10 Million Identities on a Single Machine

Xiang An, XuHan Zhu, Yang Xiao, Lan Wu, Ming Zhang, Yuan Gao, Bin Qin, Debing Zhang, Ying Fu, Jiankang Deng

Training face recognition models with millions of identities is challenging because classifier storage, logit memory, and computation grow linearly with the number of classes, eventually making full softmax impractical even when the backbone itself fits comfortably in memory. We present Partial FC (PFC), a scalable approximation to large-class softmax that preserves every positive class center while activating only a sampled subset of negative centers in each mini-batch. This asymmetric treatment retains every target term while avoiding exhaustive interaction with millions of mostly uninformative negatives. Our distributed implementation partitions the classifier across GPUs and samples within each owned shard, so sampling reduces local matrix multiplication and logit storage while sharding eliminates class-gradient synchronization across workers. Together, these properties reduce GPU-resident classifier memory, computation, and class-dependent communication without feature-based hard-negative retrieval. End-to-end system benchmarks demonstrate efficient scaling to massive class spaces, including 64 million classes on a single eight-GPU machine. Across large-scale face-recognition datasets, moderate sampling rates maintain competitive recognition accuracy while substantially improving training efficiency. Our best PFC configurations achieve 97.2\% TAR on IJB-C at FAR $=10^{-4}$ and 94.0\% TAR on ICCV21-MFR at FAR $=10^{-6}$. Beyond clean training data, PFC is robust to inter-class conflicts, label noise, and long-tailed identity distributions: under 40\% label noise, PFC with conflict filtering raises ICCV21-MFR TAR from 43.9\% to 80.2\%, while on long-tailed data PFC improves TAR from 87.4\% to 92.0\%. These results establish positive-preserving negative sampling as an effective foundation for scalable, accurate, and robust identity classification.

arXiv

Are LLMs Safe Beyond Text: Do Emojis Expose Gaps in Safety Evaluation

M P V S Gopinadh

Safety evaluations of large language models (LLMs) predominantly rely on text-based adversarial prompts, potentially overlooking vulnerabilities arising from alternative input representations. This work examines emoji-augmented prompts as a test case for this gap, evaluating 50 prompts across four open-source LLMs (Mistral 7B, Qwen 2 7B, Gemma 2 9B, Llama 3 8B). Results show substantial variation in robustness: Gemma 2 9B and Mistral 7B exhibit non-zero success rates (10%), Llama 3 8B 6%, while Qwen 2 7B shows complete resistance (0% success rate). A chi-square test ($χ^2 = 32.94, p < 0.001$) confirms significant differences in outcome distributions. These findings indicate that robustness is sensitive to input representation, and that evaluations restricted to standard text prompts may underrepresent model vulnerabilities.

arXiv

Language models suffer from a curse of ambiguity

Nicolas Zucchet, Hyun Dong Lee, Scott Linderman

Large language models increasingly rely on sampling as a driver of their own improvement, making the fidelity of their learned distributions more critical than ever. Yet, not all distributions are equally easy to learn. In this work, we identify a curse of ambiguity: in large language models, and more broadly in all neural networks that produce discrete probability distributions, the more ambiguous a next-token distribution is, the harder it is to learn accurately. Through an extensive theoretical analysis, we trace this curse to architectural and learning roots. More ambiguous distributions require more capacity to be stored, larger embeddings to be represented, more steps to be fitted, and amplify token-sampling noise. We validate these findings on synthetic tasks with controlled ground truth and observe the same signatures in language models trained on real data. Our results provide a new perspective on the statistical capabilities of large language models and a practical framework for when to trust their output distribution.

arXiv

Semantic Space of Parts of Speech

Jiří Milička, Ivan Kraus, Arnold Stanovský, Anna Vysloužilová, Barbora Štěpánková, Lenka Fárová, Vojtěch Cink, Šárka Dohnalová

Parts of speech categorization is understood in the European linguistic tradition as crisp categorization, which is also reflected in corpus linguistics, where each disambiguated token is assigned exactly one POS. However, the assigned categories are largely determined by arbitrary decisions distilled into annotation manuals. Since some words stand between parts of speech in their semantics or typical syntax, and some parts of speech are closer to each other than others, POS categorization seems inherently fuzzy. We analyze this fuzziness using word2vec embeddings, training a neural network to reduce their high dimensionality to three dimensions relevant for determining parts of speech. This creates a three-dimensional space onto which we map several thousand words, revealing which are prototypical and which lie on the boundaries, and visualizing relationships between parts of speech. The study uses Universal Dependencies POS tags for French, Czech, Finnish, Russian, and English.

arXiv

Gated Against One Model, Open to the Next: Option-Only Solvability in Legal Multiple-Choice Benchmarks

Volodymyr Ovcharov

Multiple-choice benchmarks are graded on whether a model picks the right option, not on whether it needed the question. Measuring that gap takes care: a model answering A to most items scores above chance wherever the key sits at A, and reads as recognition when it is not. We measure it on UA-JudgeExam: 11,990 four-option items with official keys, published by Ukraine's Higher Qualification Commission of Judges. Shown the options and no question, Claude Haiku 4.5 scores 0.383 against chance, and the leak is concentrated: 11.8% of items are answered blind on all eight option orders, against 0.2 items expected by chance. It is not quotation: search over 280,059 editions of Ukrainian legislation recovers 0.128. Gating those out retains 8,128 items, on which the gating model itself now scores 0.204, and GPT-5.6, which took no part in the selection, still answers 0.515 of them with the question hidden. Scoring twelve held-out models on the whole set and subtracting each one's answer-position habit, only two keep an excess: GPT-5.6 at +0.265, Sonnet 4.6 at +0.081. Without it the ranking misleads: Llama 3.1 8B scores 0.292 blind, above every model but those two, purely by answering A to 92% of items. The gate does select something real: on the items it rejected, eleven of twelve models score 0.518-0.789, every interval clear of what the same model scores on the items it kept. But that signal is one model's, and filtering on it does not transfer upward. Neither is visible on a 400-item sample, where nine models read as "statistically at chance". Rewriting distractors instead overshoots to 0.168, below chance and as exploitable. The same probe on LEXam returns chance: every option there points into the stem, none longer than 33 characters. Item format decides whether the problem can arise; capability decides how much is extracted. We release the corpus, the predictions and the harness.

arXiv

Large Language Model Assisted Operational Monitoring for Battery Energy Storage System Integrated Power Distribution Networks

Azmeer Akhtar, Md Fazley Rafy, Anurag K. Srivastava

Battery energy storage systems (BESS) are increasingly used in distribution networks for voltage regulation and demand response, which increases the volume and complexity of operational telemetry available to grid operators. This paper presents an AI-enabled monitoring framework that connects a large language model (LLM) interface with a structured telemetry database for BESS-integrated distribution system analysis. Operator questions are submitted in natural language and translated into validated SQL queries using predefined database schema information and approved KPI views. Retrieved measurements, including bus voltages, state of charge, active power, and reactive power, are evaluated against engineering constraints for voltage limits, BESS operation, and demand response tracking. The framework is validated using hardware-in-the-loop co-simulation data from a BESS-equipped distribution feeder operating under reactive power-based voltage control and price-driven demand response. Case studies show that the framework generates valid database queries, identifies repeated voltage violations, detects reactive power overshoot, and evaluates active-power tracking performance. The results show that LLM-assisted monitoring can connect structured grid telemetry with automated engineering assessment for BESS operation analysis.

arXiv

The Machine's Internal Clock: Do LLMs Share Human Temporal Illusions?

Catherine Bao, Vivek Srikumar

Human perception of time is subjective. Well-documented temporal illusions show that the brain relies on context and relational cues for judging duration instead of tracking elapsed time directly. Prior studies established these effects with visual and auditory stimuli. Existing LLM evaluations of temporal perception focus on estimating event durations or multi-step temporal reasoning. In this work, we investigate whether written narratives alone can evoke human temporal illusions, using a new benchmark of 6,684 narrative pairs spanning five illusions. We find that human readers (60 participants) prefer expected scenarios in only two of the five illusions, those where the manipulation is directly visible in text rather than requiring readers to internally simulate duration. We evaluate 14 LLMs on the same benchmark. Surprisingly, we find that models pick the literature-predicted scenario across four of the five illusions, diverging from human behavior. Reasoning traces show that ~70% of responses explicitly evoke psychology research, suggesting that this alignment is consistent with retrieval of published findings rather than human-like temporal biases.

arXiv

When AI Rewrites, Classifiers Relax: Uncertainty-Aware Sentiment Analysis on Sarcastic and AI-Paraphrased Social Text

Shresth Shroff

Sentiment classifiers are increasingly applied to social media content that is either sarcastic or AI-generated --- two distributional regimes where standard evaluations offer little guidance. We present a three-part empirical study of sentiment classifier behaviour under these conditions. First, we find that confidence scores on sarcastic text are significantly lower than on non-sarcastic text (Mann--Whitney $p = 2 \times 10^{-6}$), confirming that classifiers sense their own uncertainty on ironic content even without explicit uncertainty modelling. Second, and counterintuitively, we show that sentiment classifiers achieve higher accuracy on AI-paraphrased reviews than on the original human-authored text (RoBERTa: $+5.8$ pp for Qwen3.5-4B paraphrases, $+3.7$ pp for Gemma4-E4B), revealing a cross-domain stylistic alignment effect: AI paraphrases remove distributional noise that confounds Twitter-trained classifiers, producing cleaner, more prototypical sentiment text. Third, we demonstrate that a lightweight abstention wrapper --- flagging the $14\%$ of inputs with confidence below $0.6$ --- improves accuracy from 82.2\% to 88.9\% ($+6.7$ pp) on the retained set. We further compare Semantic Entropy and MC-Dropout-style disagreement as uncertainty signals and find near-identical AUROC ($0.650$ vs.\ $0.646$) on sarcastic text, suggesting that for short social media inputs, both methods are interchangeable. Our results motivate a shift from confident single-label prediction to uncertainty-aware abstention in high-stakes sentiment applications such as mental health flagging and content moderation.

arXiv

Logical Embeddings for Argument Analysis

Leander Heldring, Santiago Torres

We propose a new framework for machine-learning-oriented argument analysis tasks. Our proposal involves replacing traditional contextualized word embeddings used in most NLP tasks with logical embeddings, an alternative encoding that directly exploits argumentation structures. In essence, logical embeddings encapsulate the logical semantics of an argument, allowing for a better representation of its meaning. Supporting these embeddings is a mathematical logic-based similarity measure that offers a transparent notion of proximity and is guaranteed to satisfy several desirable theoretical properties that current cosine similarity-based contextualized word embeddings cannot assure. This similarity measure induces a positive semi-definite kernel on the set of arguments, enabling us to uniquely define logical embeddings using the theory of Reproducing Kernel Hilbert Spaces (RKHS). Moreover, we prove that this encoding is optimal, in the sense that no logical information is lost in the process. As with other RKHS applications, logical embeddings can be used in numerous supervised and unsupervised tasks. We provide an implementation of the method and aim to test it against literature benchmarks. Additionally, we demonstrate that logical embeddings outperform most standard embedding methods on a classification task.

arXiv

When Do Concepts Become Functionally Sufficient During Language-Model Training?

Raphael Bernas, Paul G. Chevalier, Fanny Jourdan, Céline Hudelot

Understanding a model and its learning mechanisms in depth requires identifying when its internal structures become useful, rather than simply looking at the final state. We study this through concept dynamics: at each layer and checkpoint, we decompose activations, select sparse soft masks, and inject masked reconstructions into the model. Concept analysis is therefore tested functionally: a mask is useful only insofar as it preserves a target under intervention. We compare sufficiency for activation reconstruction, linear decodability, true downstream preservation, and checkpoint transfer under learned alignment. The framework treats decomposition assumptions as hypotheses rather than interpretability guarantees, monitoring functional sufficiency across checkpoints and source-to-final reconstructability under learned alignment. At the shared fixed-penalty operating point across seven models, downstream masks retain substantially less soft mass than reconstruction masks; predictive-distribution shifts remain small.

arXiv

VTInstructor: Visual Trajectory Prompting for Navigation Instruction Generation in Continuous Environments

Haolin Yang, Yuxing Long, Zihan Yang, Hao Dong

Navigation instruction generation from ego-centric RGB video in continuous environments is an important yet challenging task for human-robot interaction and scalable dataset construction. Prior instruction generators assume discrete viewpoint graphs with panoramic observations, where trajectory structure is explicit; in continuous environments, however, the agent receives only a dense RGB stream, making trajectory cues difficult to recover. We propose VTInstructor, the first VLN instruction generation framework for continuous environments. Our key idea is to convert implicit trajectory geometry into explicit visual trajectory prompts: EDTC condenses long RGB trajectories into navigation-critical keyframes, VTP overlays path, turn, and goal cues onto these anchors, VTMod injects the resulting trajectory signals into the visual encoder, and VT-GRPO further calibrates this spatial injection during training, all without requiring a navigation graph, pre-built map, or scene reconstruction. On the challenging R2R-CE and RxR-CE Val Unseen benchmarks, VTInstructor sets a new state of the art across all standard NLG metrics, surpassing the strongest baseline by +0.357 CIDEr and +0.109 CIDEr, respectively. Beyond automatic metrics, VTInstructor-generated instructions raise a frozen follower's success rate to 63.3%, a +14.7 percentage-point gain over the best competing instruction source, and provide consistent data augmentation gains of +3 SR points on downstream navigation tasks.

继续浏览

更多 AI 顶会论文

30+ 个顶会，按会议、年份和主题浏览。

浏览全部会议

进入任一会议和年份，查看完整论文列表。

全部会议 →

搜索论文

按标题、作者或关键词，在 21 万+ 篇论文里搜索。

搜索论文 →