Mohamed Elhoseiny

Guocheng Qian, Yuchen Li, Houwen Peng et al.

PointNet++ is one of the most influential neural architectures for point cloud understanding. Although the accuracy of PointNet++ has been largely surpassed by recent networks such as PointMLP and Point Transformer, we find that a large portion of the performance gain is due to improved training strategies, i.e. data augmentation and optimization techniques, and increased model sizes rather than architectural innovations. Thus, the full potential of PointNet++ has yet to be explored. In this work, we revisit the classical PointNet++ through a systematic study of model training and scaling strategies, and offer two major contributions. First, we propose a set of improved training strategies that significantly improve PointNet++ performance. For example, we show that, without any change in architecture, the overall accuracy (OA) of PointNet++ on ScanObjectNN object classification can be raised from 77.9% to 86.1%, even outperforming state-of-the-art PointMLP. Second, we introduce an inverted residual bottleneck design and separable MLPs into PointNet++ to enable efficient and effective model scaling and propose PointNeXt, the next version of PointNets. PointNeXt can be flexibly scaled up and outperforms state-of-the-art methods on both 3D classification and segmentation tasks. For classification, PointNeXt reaches an overall accuracy of 87.7 on ScanObjectNN, surpassing PointMLP by 2.3%, while being 10x faster in inference. For semantic segmentation, PointNeXt establishes a new state-of-the-art performance with 74.9% mean IoU on S3DIS (6-fold cross-validation), being superior to the recent Point Transformer. The code and models are available at https://github.com/guochengqian/pointnext.

Xingyu Qiu, Yuqian Fu, Jiawei Geng et al.

Cross-domain few-shot object detection (CD-FSOD) remains a challenging problem for existing object detectors and few-shot learning approaches, particularly when generalizing across distinct domains. As part of NTIRE 2026, we hosted the second CD-FSOD Challenge to systematically evaluate and promote progress in detecting objects in unseen target domains under limited annotation conditions. The challenge received strong community interest, with 128 registered participants and a total of 696 submissions. Among them, 31 teams actively participated, and 19 teams submitted valid final results. Participants explored a wide range of strategies, introducing innovative methods that push the performance frontier under both open-source and closed-source tracks. This report presents a detailed overview of the NTIRE 2026 CD-FSOD Challenge, including a summary of the submitted approaches and an analysis of the final results across all participating teams. Challenge Codes: https://github.com/ohMargin/NTIRE2026_CDFSOD.

Jun Chen, Ming Hu, Darren J. Coker et al. · mit

Monitoring animal behavior can facilitate conservation efforts by providing key insights into wildlife health, population status, and ecosystem function. Automatic recognition of animals and their behaviors is critical for capitalizing on the large unlabeled datasets generated by modern video devices and for accelerating monitoring efforts at scale. However, the development of automated recognition systems is currently hindered by a lack of appropriately labeled datasets. Existing video datasets 1) do not classify animals according to established biological taxonomies; 2) are too small to facilitate large-scale behavioral studies and are often limited to a single species; and 3) do not feature temporally localized annotations and therefore do not facilitate localization of targeted behaviors within longer video sequences. Thus, we propose MammalNet, a new large-scale animal behavior dataset with taxonomy-guided annotations of mammals and their common behaviors. MammalNet contains over 18K videos totaling 539 hours, which is ~10 times larger than the largest existing animal behavior dataset. It covers 17 orders, 69 families, and 173 mammal categories for animal categorization and captures 12 high-level animal behaviors that received focus in previous animal behavior studies. We establish three benchmarks on MammalNet: standard animal and behavior recognition, compositional low-shot animal and behavior recognition, and behavior detection. Our dataset and code have been made available at: https://mammal-net.github.io.

Youssef Mohamed, Kenneth Ward Church, Mohamed Elhoseiny

We present P-Topics (Perception Topics) modeling, a novel problem for understanding how images are perceived affectively and across cultures. The goal is to (1) discover and model the different perception experiences in a dataset of images and captions, where each experience is defined by an objective factual and a subjective affective aspect, and (2) associate images to their relevant perception experiences. We introduce **PercepT** (**Percep**tion topic **T**ransformer), a two-stage architecture that tackles P-Topics modeling. In the formation stage, percepT discovers *P-Topics* as visual-textual clusters using an unsupervised training objective, and dynamically selects the number of clusters to match the perceptual richness of the dataset. In the mapping stage, it learns *P-Topic mapping functions* via attention pooling to associate images to their respective clusters. On ArtELingo, PercepT achieves a silhouette score of **0.97** compared to **0.37** from the closest baseline reflecting better perceptual clusters. PercepT also achieves an AUC score of **0.94** compared to **0.77** showing better mapping to perceptual clusters. Human evaluation confirms that PercepT captures semantically meaningful perception experiences and significantly outperforms existing methods. Our implementation will be made public.

Deyao Zhu, Yuhui Wang, Jürgen Schmidhuber et al.

Offline RL methods have been shown to reduce the need for environment interaction by training agents using offline collected episodes. However, these methods typically require action information to be logged during data collection, which can be difficult or even impossible in some practical cases. In this paper, we investigate the potential of using action-free offline datasets to improve online reinforcement learning, name this problem Reinforcement Learning with Action-Free Offline Pretraining (AFP-RL). We introduce Action-Free Guide (AF-Guide), a method that guides online training by extracting knowledge from action-free offline datasets. AF-Guide consists of an Action-Free Decision Transformer (AFDT) implementing a variant of Upside-Down Reinforcement Learning. It learns to plan the next states from the offline dataset, and a Guided Soft Actor-Critic (Guided SAC) that learns online with guidance from AFDT. Experimental results show that AF-Guide can improve sample efficiency and performance in online training thanks to the knowledge from the action-free offline dataset. Code is available at https://github.com/Vision-CAIR/AF-Guide.

Deyao Zhu, Jun Chen, Kilichbek Haydarov et al.

Asking insightful questions is crucial for acquiring knowledge and expanding our understanding of the world. However, the importance of questioning has been largely overlooked in AI research, where models have been primarily developed to answer questions. With the recent advancements of large language models (LLMs) like ChatGPT, we discover their capability to ask high-quality questions when provided with a suitable prompt. This discovery presents a new opportunity to develop an automatic questioning system. In this paper, we introduce ChatCaptioner, a novel automatic-questioning method deployed in image captioning. Here, ChatGPT is prompted to ask a series of informative questions about images to BLIP-2, a strong vision question-answering model. By keeping acquiring new visual information from BLIP-2's answers, ChatCaptioner is able to generate more enriched image descriptions. We conduct human-subject evaluations on common image caption datasets such as COCO, Conceptual Caption, and WikiArt, and compare ChatCaptioner with BLIP-2 as well as ground truth. Our results demonstrate that ChatCaptioner's captions are significantly more informative, receiving three times as many votes from human evaluators for providing the most image information. Besides, ChatCaptioner identifies 53% more objects within the image than BLIP-2 alone measured by WordNet synset matching. Code is available at https://github.com/Vision-CAIR/ChatCaptioner

Abduallah Mohamed, Deyao Zhu, Warren Vu et al.

Best-of-N (BoN) Average Displacement Error (ADE)/ Final Displacement Error (FDE) is the most used metric for evaluating trajectory prediction models. Yet, the BoN does not quantify the whole generated samples, resulting in an incomplete view of the model's prediction quality and performance. We propose a new metric, Average Mahalanobis Distance (AMD) to tackle this issue. AMD is a metric that quantifies how close the whole generated samples are to the ground truth. We also introduce the Average Maximum Eigenvalue (AMV) metric that quantifies the overall spread of the predictions. Our metrics are validated empirically by showing that the ADE/FDE is not sensitive to distribution shifts, giving a biased sense of accuracy, unlike the AMD/AMV metrics. We introduce the usage of Implicit Maximum Likelihood Estimation (IMLE) as a replacement for traditional generative models to train our model, Social-Implicit. IMLE training mechanism aligns with AMD/AMV objective of predicting trajectories that are close to the ground truth with a tight spread. Social-Implicit is a memory efficient deep model with only 5.8K parameters that runs in real time of about 580Hz and achieves competitive results. Interactive demo of the problem can be seen at https://www.abduallahmohamed.com/social-implicit-amdamv-adefde-demo . Code is available at https://github.com/abduallahmohamed/Social-Implicit .

Runjia Li, Shuyang Sun, Mohamed Elhoseiny et al. · oxford

This paper presents OxfordTVG-HIC (Humorous Image Captions), a large-scale dataset for humour generation and understanding. Humour is an abstract, subjective, and context-dependent cognitive construct involving several cognitive factors, making it a challenging task to generate and interpret. Hence, humour generation and understanding can serve as a new task for evaluating the ability of deep-learning methods to process abstract and subjective information. Due to the scarcity of data, humour-related generation tasks such as captioning remain under-explored. To address this gap, OxfordTVG-HIC offers approximately 2.9M image-text pairs with humour scores to train a generalizable humour captioning model. Contrary to existing captioning datasets, OxfordTVG-HIC features a wide range of emotional and semantic diversity resulting in out-of-context examples that are particularly conducive to generating humour. Moreover, OxfordTVG-HIC is curated devoid of offensive content. We also show how OxfordTVG-HIC can be leveraged for evaluating the humour of a generated text. Through explainability analysis of the trained models, we identify the visual and linguistic cues influential for evoking humour prediction (and generation). We observe qualitatively that these cues are aligned with the benign violation theory of humour in cognitive psychology.

Deyao Zhu, Li Erran Li, Mohamed Elhoseiny

Reinforcement Learning (RL) methods are typically applied directly in environments to learn policies. In some complex environments with continuous state-action spaces, sparse rewards, and/or long temporal horizons, learning a good policy in the original environments can be difficult. Focusing on the offline RL setting, we aim to build a simple and discrete world model that abstracts the original environment. RL methods are applied to our world model instead of the environment data for simplified policy learning. Our world model, dubbed Value Memory Graph (VMG), is designed as a directed-graph-based Markov decision process (MDP) of which vertices and directed edges represent graph states and graph actions, separately. As state-action spaces of VMG are finite and relatively small compared to the original environment, we can directly apply the value iteration algorithm on VMG to estimate graph state values and figure out the best graph actions. VMG is trained from and built on the offline RL dataset. Together with an action translator that converts the abstract graph actions in VMG to real actions in the original environment, VMG controls agents to maximize episode returns. Our experiments on the D4RL benchmark show that VMG can outperform state-of-the-art offline RL methods in several goal-oriented tasks, especially when environments have sparse rewards and long temporal horizons. Code is available at https://github.com/TsuTikgiau/ValueMemoryGraph

Yousef A. Radwan, Xuhui Liu, Kilichbek Haydarov et al.

Large language models (LLMs) have emerged as powerful representation learners whose internal features increasingly align with human cognition. We study whether modern LLMs can serve as a lens for understanding neural representations in the human brain, focusing on emotional valence in EEG. We first build a one-dimensional valence direction, the V-axis, from modern LLMs using only nine emotion-evocative sentences. We validate it through zero-shot transfer to sentiment benchmarks and cross-model consistency across fourteen LLMs. We then show that this LLM-derived direction maps onto human neural activity. On a public EEG cohort of 123 subjects watching affective videos, a single linear projection on EEG features tracks the V-axis position of each stimulus. Moreover, 36 EEG emotion classifiers trained without exposure to the V-axis spontaneously rediscover the same direction in their internal representations, suggesting that the same valence structure emerges in both language models and human electrophysiology. Yet this convergence does not provide an effective training signal. We test twenty-five alignment strategies, including knowledge distillation, representational similarity, contrastive, and topographic losses; none improve decoding, and sixteen significantly reduce accuracy. We formalize this result as the saturation regularity: once task labels alone drive a brain-decoding network onto the target direction, additional supervision mainly distorts an already-saturated basin, while the load-bearing within-class residual receives little useful gradient. This regularity also indicates where improvement should come from: the residual subspace unreachable by supervision. Motivated by this insight, we ensemble across residual diversity rather than supervising the basin, improving balanced accuracy by 10.5% over the prior best on FACED, with the same effect replicated on SEED-V.

Eslam Mohamed Bakr, Pengzhan Sun, Li Erran Li et al.

Most pre-trained learning systems are known to suffer from bias, which typically emerges from the data, the model, or both. Measuring and quantifying bias and its sources is a challenging task and has been extensively studied in image captioning. Despite the significant effort in this direction, we observed that existing metrics lack consistency in the inclusion of the visual signal. In this paper, we introduce a new bias assessment metric, dubbed $ImageCaptioner^2$, for image captioning. Instead of measuring the absolute bias in the model or the data, $ImageCaptioner^2$ pay more attention to the bias introduced by the model w.r.t the data bias, termed bias amplification. Unlike the existing methods, which only evaluate the image captioning algorithms based on the generated captions only, $ImageCaptioner^2$ incorporates the image while measuring the bias. In addition, we design a formulation for measuring the bias of generated captions as prompt-based image captioning instead of using language classifiers. Finally, we apply our $ImageCaptioner^2$ metric across 11 different image captioning architectures on three different datasets, i.e., MS-COCO caption dataset, Artemis V1, and Artemis V2, and on three different protected attributes, i.e., gender, race, and emotions. Consequently, we verify the effectiveness of our $ImageCaptioner^2$ metric by proposing AnonymousBench, which is a novel human evaluation paradigm for bias metrics. Our metric shows significant superiority over the recent bias metric; LIC, in terms of human alignment, where the correlation scores are 80% and 54% for our metric and LIC, respectively. The code is available at https://eslambakr.github.io/imagecaptioner2.github.io/.

Eslam Mohamed Bakr, Yasmeen Alsaedy, Mohamed Elhoseiny

The 3D visual grounding task has been explored with visual and language streams comprehending referential language to identify target objects in 3D scenes. However, most existing methods devote the visual stream to capturing the 3D visual clues using off-the-shelf point clouds encoders. The main question we address in this paper is "can we consolidate the 3D visual stream by 2D clues synthesized from point clouds and efficiently utilize them in training and testing?". The main idea is to assist the 3D encoder by incorporating rich 2D object representations without requiring extra 2D inputs. To this end, we leverage 2D clues, synthetically generated from 3D point clouds, and empirically show their aptitude to boost the quality of the learned visual representations. We validate our approach through comprehensive experiments on Nr3D, Sr3D, and ScanRefer datasets and show consistent performance gains compared to existing methods. Our proposed module, dubbed as Look Around and Refer (LAR), significantly outperforms the state-of-the-art 3D visual grounding techniques on three benchmarks, i.e., Nr3D, Sr3D, and ScanRefer. The code is available at https://eslambakr.github.io/LAR.github.io/.

Jun Chen, Ming Hu, Boyang Li et al.

Self-supervised learning for computer vision has achieved tremendous progress and improved many downstream vision tasks such as image classification, semantic segmentation, and object detection. Among these, generative self-supervised vision learning approaches such as MAE and BEiT show promising performance. However, their global masked reconstruction mechanism is computationally demanding. To address this issue, we propose local masked reconstruction (LoMaR), a simple yet effective approach that performs masked reconstruction within a small window of 7$\times$7 patches on a simple Transformer encoder, improving the trade-off between efficiency and accuracy compared to global masked reconstruction over the entire image. Extensive experiments show that LoMaR reaches 84.1% top-1 accuracy on ImageNet-1K classification, outperforming MAE by 0.5%. After finetuning the pretrained LoMaR on 384$\times$384 images, it can reach 85.4% top-1 accuracy, surpassing MAE by 0.6%. On MS COCO, LoMaR outperforms MAE by 0.5 $\text{AP}^\text{box}$ on object detection and 0.5 $\text{AP}^\text{mask}$ on instance segmentation. LoMaR is especially more computation-efficient on pretraining high-resolution images, e.g., it is 3.1$\times$ faster than MAE with 0.2% higher classification accuracy on pretraining 448$\times$448 images. This local masked reconstruction learning mechanism can be easily integrated into any other generative self-supervised learning approach. Our code is publicly available in https://github.com/junchen14/LoMaR.

Jun Chen, Deyao Zhu, Kilichbek Haydarov et al.

Video captioning aims to convey dynamic scenes from videos using natural language, facilitating the understanding of spatiotemporal information within our environment. Although there have been recent advances, generating detailed and enriched video descriptions continues to be a substantial challenge. In this work, we introduce Video ChatCaptioner, an innovative approach for creating more comprehensive spatiotemporal video descriptions. Our method employs a ChatGPT model as a controller, specifically designed to select frames for posing video content-driven questions. Subsequently, a robust algorithm is utilized to answer these visual queries. This question-answer framework effectively uncovers intricate video details and shows promise as a method for enhancing video content. Following multiple conversational rounds, ChatGPT can summarize enriched video content based on previous conversations. We qualitatively demonstrate that our Video ChatCaptioner can generate captions containing more visual details about the videos. The code is publicly available at https://github.com/Vision-CAIR/ChatCaptioner

Xiaoqian Shen, Xiang Li, Mohamed Elhoseiny

Video generation remains a challenging task due to spatiotemporal complexity and the requirement of synthesizing diverse motions with temporal consistency. Previous works attempt to generate videos in arbitrary lengths either in an autoregressive manner or regarding time as a continuous signal. However, they struggle to synthesize detailed and diverse motions with temporal coherence and tend to generate repetitive scenes after a few time steps. In this work, we argue that a single time-agnostic latent vector of style-based generator is insufficient to model various and temporally-consistent motions. Hence, we introduce additional time-dependent motion styles to model diverse motion patterns. In addition, a Motion Style Attention modulation mechanism, dubbed as MoStAtt, is proposed to augment frames with vivid dynamics for each specific scale (i.e., layer), which assigns attention score for each motion style w.r.t deconvolution filter weights in the target synthesis layer and softly attends different motion styles for weight modulation. Experimental results show our model achieves state-of-the-art performance on four unconditional $256^2$ video synthesis benchmarks trained with only 3 frames per clip and produces better qualitative results with respect to dynamic motions. Code and videos have been made available at https://github.com/xiaoqian-shen/MoStGAN-V.

Jun Chen, Deyao Zhu, Guocheng Qian et al. · amazon-science

Semantic segmentation is a crucial task in computer vision that involves segmenting images into semantically meaningful regions at the pixel level. However, existing approaches often rely on expensive human annotations as supervision for model training, limiting their scalability to large, unlabeled datasets. To address this challenge, we present ZeroSeg, a novel method that leverages the existing pretrained vision-language (VL) model (e.g. CLIP) to train open-vocabulary zero-shot semantic segmentation models. Although acquired extensive knowledge of visual concepts, it is non-trivial to exploit knowledge from these VL models to the task of semantic segmentation, as they are usually trained at an image level. ZeroSeg overcomes this by distilling the visual concepts learned by VL models into a set of segment tokens, each summarizing a localized region of the target image. We evaluate ZeroSeg on multiple popular segmentation benchmarks, including PASCAL VOC 2012, PASCAL Context, and COCO, in a zero-shot manner (i.e., no training or adaption on target segmentation datasets). Our approach achieves state-of-the-art performance when compared to other zero-shot segmentation methods under the same training data, while also performing competitively compared to strongly supervised methods. Finally, we also demonstrated the effectiveness of ZeroSeg on open-vocabulary segmentation, through both human studies and qualitative visualizations.

Sanjoy Chowdhury, Sayan Nag, Subhrajyoti Dasgupta et al.

Leveraging Large Language Models' remarkable proficiency in text-based tasks, recent works on Multi-modal LLMs (MLLMs) extend them to other modalities like vision and audio. However, the progress in these directions has been mostly focused on tasks that only require a coarse-grained understanding of the audio-visual semantics. We present Meerkat, an audio-visual LLM equipped with a fine-grained understanding of image and audio both spatially and temporally. With a new modality alignment module based on optimal transport and a cross-attention module that enforces audio-visual consistency, Meerkat can tackle challenging tasks such as audio referred image grounding, image guided audio temporal localization, and audio-visual fact-checking. Moreover, we carefully curate a large dataset AVFIT that comprises 3M instruction tuning samples collected from open-source datasets, and introduce MeerkatBench that unifies five challenging audio-visual tasks. We achieve state-of-the-art performance on all these downstream tasks with a relative improvement of up to 37.12%.

Eslam Abdelrahman, Mohamed Ayman, Mahmoud Ahmed et al.

3D visual grounding is the ability to localize objects in 3D scenes conditioned by utterances. Most existing methods devote the referring head to localize the referred object directly, causing failure in complex scenarios. In addition, it does not illustrate how and why the network reaches the final decision. In this paper, we address this question Can we design an interpretable 3D visual grounding framework that has the potential to mimic the human perception system?. To this end, we formulate the 3D visual grounding problem as a sequence-to-sequence Seq2Seq task by first predicting a chain of anchors and then the final target. Interpretability not only improves the overall performance but also helps us identify failure cases. Following the chain of thoughts approach enables us to decompose the referring task into interpretable intermediate steps, boosting the performance and making our framework extremely data-efficient. Moreover, our proposed framework can be easily integrated into any existing architecture. We validate our approach through comprehensive experiments on the Nr3D, Sr3D, and Scanrefer benchmarks and show consistent performance gains compared to existing methods without requiring manually annotated data. Furthermore, our proposed framework, dubbed CoT3DRef, is significantly data-efficient, whereas on the Sr3D dataset, when trained only on 10% of the data, we match the SOTA performance that trained on the entire data. The code is available at https:eslambakr.github.io/cot3dref.github.io/.

Deyao Zhu, Jun Chen, Xiaoqian Shen et al.

The recent GPT-4 has demonstrated extraordinary multi-modal abilities, such as directly generating websites from handwritten text and identifying humorous elements within images. These features are rarely observed in previous vision-language models. However, the technical details behind GPT-4 continue to remain undisclosed. We believe that the enhanced multi-modal generation capabilities of GPT-4 stem from the utilization of sophisticated large language models (LLM). To examine this phenomenon, we present MiniGPT-4, which aligns a frozen visual encoder with a frozen advanced LLM, Vicuna, using one projection layer. Our work, for the first time, uncovers that properly aligning the visual features with an advanced large language model can possess numerous advanced multi-modal abilities demonstrated by GPT-4, such as detailed image description generation and website creation from hand-drawn drafts. Furthermore, we also observe other emerging capabilities in MiniGPT-4, including writing stories and poems inspired by given images, teaching users how to cook based on food photos, and so on. In our experiment, we found that the model trained on short image caption pairs could produce unnatural language outputs (e.g., repetition and fragmentation). To address this problem, we curate a detailed image description dataset in the second stage to finetune the model, which consequently improves the model's generation reliability and overall usability. Our code, pre-trained model, and collected dataset are available at https://minigpt-4.github.io/.

Jinjie Mai, Jun Chen, Bing Li et al.

Embodied AI focuses on the study and development of intelligent systems that possess a physical or virtual embodiment (i.e. robots) and are able to dynamically interact with their environment. Memory and control are the two essential parts of an embodied system and usually require separate frameworks to model each of them. In this paper, we propose a novel and generalizable framework called LLM-Brain: using Large-scale Language Model as a robotic brain to unify egocentric memory and control. The LLM-Brain framework integrates multiple multimodal language models for robotic tasks, utilizing a zero-shot learning approach. All components within LLM-Brain communicate using natural language in closed-loop multi-round dialogues that encompass perception, planning, control, and memory. The core of the system is an embodied LLM to maintain egocentric memory and control the robot. We demonstrate LLM-Brain by examining two downstream tasks: active exploration and embodied question answering. The active exploration tasks require the robot to extensively explore an unknown environment within a limited number of actions. Meanwhile, the embodied question answering tasks necessitate that the robot answers questions based on observations acquired during prior explorations.

Jun Chen, Deyao Zhu, Xiaoqian Shen et al.

Large language models have shown their remarkable capabilities as a general interface for various language-related applications. Motivated by this, we target to build a unified interface for completing many vision-language tasks including image description, visual question answering, and visual grounding, among others. The challenge is to use a single model for performing diverse vision-language tasks effectively with simple multi-modal instructions. Towards this objective, we introduce MiniGPT-v2, a model that can be treated as a unified interface for better handling various vision-language tasks. We propose using unique identifiers for different tasks when training the model. These identifiers enable our model to better distinguish each task instruction effortlessly and also improve the model learning efficiency for each task. After the three-stage training, the experimental results show that MiniGPT-v2 achieves strong performance on many visual question-answering and visual grounding benchmarks compared to other vision-language generalist models. Our model and codes are available at https://minigpt-v2.github.io/

Youssef Mohamed, Mohamed Abdelfattah, Shyma Alhuwaider et al.

This paper introduces ArtELingo, a new benchmark and dataset, designed to encourage work on diversity across languages and cultures. Following ArtEmis, a collection of 80k artworks from WikiArt with 0.45M emotion labels and English-only captions, ArtELingo adds another 0.79M annotations in Arabic and Chinese, plus 4.8K in Spanish to evaluate "cultural-transfer" performance. More than 51K artworks have 5 annotations or more in 3 languages. This diversity makes it possible to study similarities and differences across languages and cultures. Further, we investigate captioning tasks, and find diversity improves the performance of baseline models. ArtELingo is publicly available at https://www.artelingo.org/ with standard splits and baseline models. We hope our work will help ease future research on multilinguality and culturally-aware AI.

Wenxuan Zhang, Philip H. S. Torr, Mohamed Elhoseiny et al.

Fine-tuning large language models (LLMs) on human preferences, typically through reinforcement learning from human feedback (RLHF), has proven successful in enhancing their capabilities. However, ensuring the safety of LLMs during fine-tuning remains a critical concern, and mitigating the potential conflicts in safety and helpfulness is costly in RLHF. To address this issue, we propose a supervised learning framework called Bi-Factorial Preference Optimization (BFPO), which re-parameterizes a joint RLHF objective of both safety and helpfulness into a single supervised learning objective. In supervised optimization, a labeling function is used to capture the global preferences ranking to balance both safety and helpfulness. To evaluate BFPO, we develop a benchmark that includes comprehensive discriminative and generative tasks for helpfulness and harmlessness. The results indicate that our method significantly outperforms existing approaches in both safety and helpfulness. Moreover, BFPO achieves the same level of safety as methods that heavily rely on human labor with less than 10\% of the computational resources and human prompting and annotation process. The training recipes can be found here: https://github.com/wx-zhang/bfpo.

Wenxuan Zhang, Paul Janson, Kai Yi et al.

Learning novel concepts, remembering previous knowledge, and adapting it to future tasks occur simultaneously throughout a human's lifetime. To model such comprehensive abilities, continual zero-shot learning (CZSL) has recently been introduced. However, most existing methods overused unseen semantic information that may not be continually accessible in realistic settings. In this paper, we address the challenge of continual zero-shot learning where unseen information is not provided during training, by leveraging generative modeling. The heart of the generative-based methods is to learn quality representations from seen classes to improve the generative understanding of the unseen visual space. Motivated by this, we introduce generalization-bound tools and provide the first theoretical explanation for the benefits of generative modeling to CZSL tasks. Guided by the theoretical analysis, we then propose our learning algorithm that employs a novel semantically guided Generative Random Walk (GRW) loss. The GRW loss augments the training by continually encouraging the model to generate realistic and characterized samples to represent the unseen space. Our algorithm achieves state-of-the-art performance on AWA1, AWA2, CUB, and SUN datasets, surpassing existing CZSL methods by 3-7\%. The code has been made available here \url{https://github.com/wx-zhang/IGCZSL}

Eslam Mohamed Bakr, Pengzhan Sun, Xiaoqian Shen et al.

In recent years, Text-to-Image (T2I) models have been extensively studied, especially with the emergence of diffusion models that achieve state-of-the-art results on T2I synthesis tasks. However, existing benchmarks heavily rely on subjective human evaluation, limiting their ability to holistically assess the model's capabilities. Furthermore, there is a significant gap between efforts in developing new T2I architectures and those in evaluation. To address this, we introduce HRS-Bench, a concrete evaluation benchmark for T2I models that is Holistic, Reliable, and Scalable. Unlike existing bench-marks that focus on limited aspects, HRS-Bench measures 13 skills that can be categorized into five major categories: accuracy, robustness, generalization, fairness, and bias. In addition, HRS-Bench covers 50 scenarios, including fashion, animals, transportation, food, and clothes. We evaluate nine recent large-scale T2I models using metrics that cover a wide range of skills. A human evaluation aligned with 95% of our evaluations on average was conducted to probe the effectiveness of HRS-Bench. Our experiments demonstrate that existing models often struggle to generate images with the desired count of objects, visual text, or grounded emotions. We hope that our benchmark help ease future text-to-image generation research. The code and data are available at https://eslambakr.github.io/hrsbench.github.io

Weiyu Ma, Yongcheng Zeng, Yan Song et al.

Reinforcement Learning (RL) has achieved impressive success in post-training Large Language Models (LLMs) and Vision-Language Models (VLMs), with on-policy algorithms such as PPO, GRPO, and REINFORCE++ serving as the dominant paradigm. However, these methods discard all collected trajectories after a single gradient update, resulting in poor sample efficiency, particularly wasteful for agentic tasks where multi-turn environment interactions are expensive. While Experience Replay drives sample efficiency in classic RL by allowing agents to reuse past trajectories and prioritize informative ones, directly applying Prioritized Experience Replay (PER) to LLMs fails. The rapid policy evolution of billion-parameter models renders stored priorities stale, causing old high-priority trajectories to dominate sampling long after they have become uninformative. We propose Freshness-Aware PER, which addresses this priority staleness problem by augmenting any PER-based priority with a multiplicative exponential age decay grounded in effective sample size analysis. To the best of our knowledge, Freshness-Aware PER is the first work to successfully apply PER to LLM/VLM reinforcement learning. We evaluate on eight multi-step agentic, reasoning, and math competition tasks with 0.5B, 3B, and 7B models. Freshness-Aware PER significantly outperforms on-policy baselines, achieving +46% on NQ Search, +367% on Sokoban, and +133% on VLM FrozenLake, while standard PER without age decay consistently degrades performance. Our code is publicly available at https://github.com/Vision-CAIR/Freshness-Aware-PER.

Liangbing Zhao, Le Zhuo, Sayak Paul et al.

Instruction-based image editing has achieved remarkable success in semantic alignment, yet state-of-the-art models frequently fail to render physically plausible results when editing involves complex causal dynamics, such as refraction or material deformation. We attribute this limitation to the dominant paradigm that treats editing as a discrete mapping between image pairs, which provides only boundary conditions and leaves transition dynamics underspecified. To address this, we reformulate physics-aware editing as predictive physical state transitions and introduce PhysicTran38K, a large-scale video-based dataset comprising 38K transition trajectories across five physical domains, constructed via a two-stage filtering and constraint-aware annotation pipeline. Building on this supervision, we propose PhysicEdit, an end-to-end framework equipped with a textual-visual dual-thinking mechanism. It combines a frozen Qwen2.5-VL for physically grounded reasoning with learnable transition queries that provide timestep-adaptive visual guidance to a diffusion backbone. Experiments show that PhysicEdit improves over Qwen-Image-Edit by 5.9% in physical realism and 10.1% in knowledge-grounded editing, setting a new state-of-the-art for open-source methods, while remaining competitive with leading proprietary models.

Youssef Mohamed, Faizan Farooq Khan, Kilichbek Haydarov et al.

Datasets that capture the connection between vision, language, and affection are limited, causing a lack of understanding of the emotional aspect of human intelligence. As a step in this direction, the ArtEmis dataset was recently introduced as a large-scale dataset of emotional reactions to images along with language explanations of these chosen emotions. We observed a significant emotional bias towards instance-rich emotions, making trained neural speakers less accurate in describing under-represented emotions. We show that collecting new data, in the same way, is not effective in mitigating this emotional bias. To remedy this problem, we propose a contrastive data collection approach to balance ArtEmis with a new complementary dataset such that a pair of similar images have contrasting emotions (one positive and one negative). We collected 260,533 instances using the proposed method, we combine them with ArtEmis, creating a second iteration of the dataset. The new combined dataset, dubbed ArtEmis v2.0, has a balanced distribution of emotions with explanations revealing more fine details in the associated painting. Our experiments show that neural speakers trained on the new dataset improve CIDEr and METEOR evaluation metrics by 20% and 7%, respectively, compared to the biased dataset. Finally, we also show that the performance per emotion of neural speakers is improved across all the emotion categories, significantly on under-represented emotions. The collected dataset and code are available at https://artemisdataset-v2.org.

Paul Janson, Wenxuan Zhang, Rahaf Aljundi et al.

With the success of pretraining techniques in representation learning, a number of continual learning methods based on pretrained models have been proposed. Some of these methods design continual learning mechanisms on the pre-trained representations and only allow minimum updates or even no updates of the backbone models during the training of continual learning. In this paper, we question whether the complexity of these models is needed to achieve good performance by comparing them to a simple baseline that we designed. We argue that the pretrained feature extractor itself can be strong enough to achieve a competitive or even better continual learning performance on Split-CIFAR100 and CoRe 50 benchmarks. To validate this, we conduct a very simple baseline that 1) use the frozen pretrained model to extract image features for every class encountered during the continual learning stage and compute their corresponding mean features on training data, and 2) predict the class of the input based on the nearest neighbor distance between test samples and mean features of the classes; i.e., Nearest Mean Classifier (NMC). This baseline is single-headed, exemplar-free, and can be task-free (by updating the means continually). This baseline achieved 88.53% on 10-Split-CIFAR-100, surpassing most state-of-the-art continual learning methods that are all initialized using the same pretrained transformer model. We hope our baseline may encourage future progress in designing learning systems that can continually add quality to the learning representations even if they started from some pretrained weights.

Mohamed Elmoghany, Liangbing Zhao, Xiaoqian Shen et al. · allen-ai

Generating long-form storytelling videos with consistent visual narratives remains a significant challenge in video synthesis. We present a novel framework, dataset, and a model that address three critical limitations: background consistency across shots, seamless multi-subject shot-to-shot transitions, and scalability to hour-long narratives. Our approach introduces a background-consistent generation pipeline that maintains visual coherence across scenes while preserving character identity and spatial relationships. We further propose a transition-aware video synthesis module that generates smooth shot transitions for complex scenarios involving multiple subjects entering or exiting frames, going beyond the single-subject limitations of prior work. To support this, we contribute with a synthetic dataset of 10,000 multi-subject transition sequences covering underrepresented dynamic scene compositions. On VBench, InfinityStory achieves the highest Background Consistency (88.94), highest Subject Consistency (82.11), and the best overall average rank (2.80), showing improved stability, smoother transitions, and better temporal coherence.

Asma Alkhaldi, Raneem Alnajim, Layan Alabdullatef et al.

Recent advancements in artificial intelligence (AI) have precipitated significant breakthroughs in healthcare, particularly in refining diagnostic procedures. However, previous studies have often been constrained to limited functionalities. This study introduces MiniGPT-Med, a vision-language model derived from large-scale language models and tailored for medical applications. MiniGPT-Med demonstrates remarkable versatility across various imaging modalities, including X-rays, CT scans, and MRIs, enhancing its utility. The model is capable of performing tasks such as medical report generation, visual question answering (VQA), and disease identification within medical imagery. Its integrated processing of both image and textual clinical data markedly improves diagnostic accuracy. Our empirical assessments confirm MiniGPT-Med's superior performance in disease grounding, medical report generation, and VQA benchmarks, representing a significant step towards reducing the gap in assisting radiology practice. Furthermore, it achieves state-of-the-art performance on medical report generation, higher than the previous best model by 19\% accuracy. MiniGPT-Med promises to become a general interface for radiology diagnoses, enhancing diagnostic efficiency across a wide range of medical imaging applications.

Kirolos Ataallah, Xiaoqian Shen, Eslam Abdelrahman et al.

Most current LLM-based models for video understanding can process videos within minutes. However, they struggle with lengthy videos due to challenges such as "noise and redundancy", as well as "memory and computation" constraints. In this paper, we present Goldfish, a methodology tailored for comprehending videos of arbitrary lengths. We also introduce the TVQA-long benchmark, specifically designed to evaluate models' capabilities in understanding long videos with questions in both vision and text content. Goldfish approaches these challenges with an efficient retrieval mechanism that initially gathers the top-k video clips relevant to the instruction before proceeding to provide the desired response. This design of the retrieval mechanism enables the Goldfish to efficiently process arbitrarily long video sequences, facilitating its application in contexts such as movies or television series. To facilitate the retrieval process, we developed MiniGPT4-Video that generates detailed descriptions for the video clips. In addressing the scarcity of benchmarks for long video evaluation, we adapted the TVQA short video benchmark for extended content analysis by aggregating questions from entire episodes, thereby shifting the evaluation from partial to full episode comprehension. We attained a 41.78% accuracy rate on the TVQA-long benchmark, surpassing previous methods by 14.94%. Our MiniGPT4-Video also shows exceptional performance in short video comprehension, exceeding existing state-of-the-art methods by 3.23%, 2.03%, 16.5% and 23.59% on the MSVD, MSRVTT, TGIF, and TVQA short video benchmarks, respectively. These results indicate that our models have significant improvements in both long and short-video understanding. Our models and code have been made publicly available at https://vision-cair.github.io/Goldfish_website/

Kai Yi, Xiaoqian Shen, Yunhao Gou et al.

The main question we address in this paper is how to scale up visual recognition of unseen classes, also known as zero-shot learning, to tens of thousands of categories as in the ImageNet-21K benchmark. At this scale, especially with many fine-grained categories included in ImageNet-21K, it is critical to learn quality visual semantic representations that are discriminative enough to recognize unseen classes and distinguish them from seen ones. We propose a \emph{H}ierarchical \emph{G}raphical knowledge \emph{R}epresentation framework for the confidence-based classification method, dubbed as HGR-Net. Our experimental results demonstrate that HGR-Net can grasp class inheritance relations by utilizing hierarchical conceptual knowledge. Our method significantly outperformed all existing techniques, boosting the performance by 7\% compared to the runner-up approach on the ImageNet-21K benchmark. We show that HGR-Net is learning-efficient in few-shot scenarios. We also analyzed our method on smaller datasets like ImageNet-21K-P, 2-hops and 3-hops, demonstrating its generalization ability. Our benchmark and code are available at https://kaiyi.me/p/hgrnet.html.

Habib Slim, Xiang Li, Yuchen Li et al.

In this work, we present 3DCoMPaT$^{++}$, a multimodal 2D/3D dataset with 160 million rendered views of more than 10 million stylized 3D shapes carefully annotated at the part-instance level, alongside matching RGB point clouds, 3D textured meshes, depth maps, and segmentation masks. 3DCoMPaT$^{++}$ covers 41 shape categories, 275 fine-grained part categories, and 293 fine-grained material classes that can be compositionally applied to parts of 3D objects. We render a subset of one million stylized shapes from four equally spaced views as well as four randomized views, leading to a total of 160 million renderings. Parts are segmented at the instance level, with coarse-grained and fine-grained semantic levels. We introduce a new task, called Grounded CoMPaT Recognition (GCR), to collectively recognize and ground compositions of materials on parts of 3D objects. Additionally, we report the outcomes of a data challenge organized at CVPR2023, showcasing the winning method's utilization of a modified PointNet$^{++}$ model trained on 6D inputs, and exploring alternative techniques for GCR enhancement. We hope our work will help ease future research on compositional 3D Vision.

Wenxuan Zhang, Paul Janson, Rahaf Aljundi et al.

Foundation models encompass an extensive knowledge base and offer remarkable transferability. However, this knowledge becomes outdated or insufficient over time. The challenge lies in continuously updating foundation models to accommodate novel information while retaining their original capabilities. Leveraging the fact that foundation models have initial knowledge on various tasks and domains, we propose a novel approach that, instead of updating all parameters equally, localizes the updates to a sparse set of parameters relevant to the task being learned. We strike a balance between efficiency and new task performance, while maintaining the transferability and generalizability of foundation models. We extensively evaluate our method on foundational vision-language models with a diverse spectrum of continual learning tasks. Our method achieves improvements on the accuracy of the newly learned tasks up to 7% while preserving the pretraining knowledge with a negligible decrease of 0.9% on a representative control set accuracy.

Chenhui Gou, Abdulwahab Felemban, Faizan Farooq Khan et al.

Large Language Model-based Vision-Language Models (LLM-based VLMs) have demonstrated impressive results in various vision-language understanding tasks. However, how well these VLMs can see image detail beyond the semantic level remains unclear. In our study, we introduce a pixel value prediction task (PVP) to explore "How Well Can Vision Language Models See Image Details?" and to assist VLMs in perceiving more details. Typically, these models comprise a frozen CLIP visual encoder, a large language model, and a connecting module. After fine-tuning VLMs on the PVP task, we find: 1) existing VLMs struggle to predict precise pixel values by only fine-tuning the connection module and LLM; and 2) prediction precision is significantly improved when the vision encoder is also adapted. Additionally, our research reveals that incorporating pixel value prediction as one of the VLM pre-training tasks and vision encoder adaptation markedly boosts VLM performance on downstream image-language understanding tasks requiring detailed image perception, such as referring image segmentation (with an average +10.19 cIoU improvement) and video game decision making (with average score improvements of +80.34 and +70.54 on two games, respectively).

Kilichbek Haydarov, Xiaoqian Shen, Avinash Madasu et al.

We introduce Affective Visual Dialog, an emotion explanation and reasoning task as a testbed for research on understanding the formation of emotions in visually grounded conversations. The task involves three skills: (1) Dialog-based Question Answering (2) Dialog-based Emotion Prediction and (3) Affective emotion explanation generation based on the dialog. Our key contribution is the collection of a large-scale dataset, dubbed AffectVisDial, consisting of 50K 10-turn visually grounded dialogs as well as concluding emotion attributions and dialog-informed textual emotion explanations, resulting in a total of 27,180 working hours. We explain our design decisions in collecting the dataset and introduce the questioner and answerer tasks that are associated with the participants in the conversation. We train and demonstrate solid Affective Visual Dialog baselines adapted from state-of-the-art models. Remarkably, the responses generated by our models show promising emotional reasoning abilities in response to visually grounded conversations. Our project page is available at https://affective-visual-dialog.github.io.

Zilyu Ye, Jinxiu Liu, Ruotian Peng et al.

Recent image generation models excel at creating high-quality images from brief captions. However, they fail to maintain consistency of multiple instances across images when encountering lengthy contexts. This inconsistency is largely due to in existing training datasets the absence of granular instance feature labeling in existing training datasets. To tackle these issues, we introduce Openstory++, a large-scale dataset combining additional instance-level annotations with both images and text. Furthermore, we develop a training methodology that emphasizes entity-centric image-text generation, ensuring that the models learn to effectively interweave visual and textual information. Specifically, Openstory++ streamlines the process of keyframe extraction from open-domain videos, employing vision-language models to generate captions that are then polished by a large language model for narrative continuity. It surpasses previous datasets by offering a more expansive open-domain resource, which incorporates automated captioning, high-resolution imagery tailored for instance count, and extensive frame sequences for temporal consistency. Additionally, we present Cohere-Bench, a pioneering benchmark framework for evaluating the image generation tasks when long multimodal context is provided, including the ability to keep the background, style, instances in the given context coherent. Compared to existing benchmarks, our work fills critical gaps in multi-modal generation, propelling the development of models that can adeptly generate and interpret complex narratives in open-domain environments. Experiments conducted within Cohere-Bench confirm the superiority of Openstory++ in nurturing high-quality visual storytelling models, enhancing their ability to address open-domain generation tasks. More details can be found at https://openstorypp.github.io/

Eslam Abdelrahman, Liangbing Zhao, Vincent Tao Hu et al.

Diffusion models break down the challenging task of generating data from high-dimensional distributions into a series of easier denoising steps. Inspired by this paradigm, we propose a novel approach that extends the diffusion framework into modality space, decomposing the complex task of RGB image generation into simpler, interpretable stages. Our method, termed ToddlerDiffusion, cascades modality-specific models, each responsible for generating an intermediate representation, such as contours, palettes, and detailed textures, ultimately culminating in a high-quality RGB image. Instead of relying on the naive LDM concatenation conditioning mechanism to connect the different stages together, we employ Schrödinger Bridge to determine the optimal transport between different modalities. Although employing a cascaded pipeline introduces more stages, which could lead to a more complex architecture, each stage is meticulously formulated for efficiency and accuracy, surpassing Stable-Diffusion (LDM) performance. Modality composition not only enhances overall performance but enables emerging proprieties such as consistent editing, interaction capabilities, high-level interpretability, and faster convergence and sampling rate. Extensive experiments on diverse datasets, including LSUN-Churches, ImageNet, CelebHQ, and LAION-Art, demonstrate the efficacy of our approach, consistently outperforming state-of-the-art methods. For instance, ToddlerDiffusion achieves notable efficiency, matching LDM performance on LSUN-Churches while operating 2$\times$ faster with a 3$\times$ smaller architecture. The project website is available at: https://toddlerdiffusion.github.io/website/

Seung Hun Han, Youssef Mohamed, Mohamed Elhoseiny

This paper presents a Multilingual Vision Large Language Model, named M-MiniGPT4. Our model exhibits strong vision-language understanding (VLU) capabilities across 11 languages. We utilize a mixture of native multilingual and translated data to push the multilingual VLU performance of the MiniGPT4 architecture. In addition, we propose a multilingual alignment training stage that uses parallel text corpora to further enhance the multilingual capabilities of our model. M-MiniGPT4 achieves 36% accuracy on the multilingual MMMU benchmark, outperforming state-of-the-art models in the same weight class, including foundation models released after the majority of this work was completed. We open-source our models, code, and translated datasets to facilitate future research in low-resource and multilingual settings.

Runze Wang, Yuqian Fu, Yu Li et al.

Vision-language-action (VLA) models have shown strong potential for generalist robot manipulation, yet they remain limited by insufficient spatial reasoning, particularly in determining where to interact in complex visual scenes. While recent efforts introduce various forms of visual planning to address this issue, existing approaches either rely on global geometric cues, symbolic intermediate representations, or externally generated visual signals, which are often weakly coupled with downstream action prediction. In this work, we revisit visual planning in VLA systems and argue that effective planning should be local, visually grounded, internally generated, and directly aligned with action. Based on this insight, we propose Afford-VLA, a unified framework that internalizes task-conditioned affordance as an explicit visual planning interface within VLA models. Concretely, we introduce learnable <AFF> tokens to query task-relevant interaction regions, decode affordance masks from multimodal features, and convert them into compact embeddings that directly condition action generation. This design enables affordance to be both generated and utilized within the VLA, forming a tightly coupled perception-action pathway. To further support this integration, we adopt a training strategy that allows the affordance pathway to be jointly optimized with action prediction, improving its effectiveness for downstream control. We evaluate our method on multiple simulation benchmarks, including LIBERO, LIBERO-Plus, and SimplerEnv, achieving consistent state-of-the-art performance, along with strong real-world results. These findings demonstrate that internalizing affordance as action-aligned visual planning provides a powerful paradigm for improving VLA systems.

Xiaoqian Shen, Min-Hung Chen, Yu-Chiang Frank Wang et al.

Grounded video question answering (GVQA) aims to localize relevant temporal segments in videos and generate accurate answers to a given question; however, large video-language models (LVLMs) exhibit limited temporal awareness. Although existing approaches based on Group Relative Policy Optimization (GRPO) attempt to improve temporal grounding, they still struggle to faithfully ground their answers in the relevant video evidence, leading to temporal mislocalization and hallucinations. In this work, we present Zoom-Zero, a coarse-to-fine framework that first localizes query-relevant segments and then temporally zooms into the most salient frames for finer-grained visual verification. Our method addresses the limits of GRPO for the GVQA task with two key innovations: (i) a zoom-in accuracy reward that validates the fidelity of temporal grounding prediction and facilitates fine-grained visual verification on grounded frames; (ii) token-selective credit assignment, which attributes rewards to the tokens responsible for temporal localization or answer generation, mitigating GRPO's issue in handling multi-faceted reward signals. Our proposed method advances grounded video question answering, improving temporal grounding by 5.2\% on NExT-GQA and 4.6\% on ReXTime, while also enhancing average answer accuracy by 2.4\%. Additionally, the coarse-to-fine zoom-in during inference further benefits long-form video understanding by preserving critical visual details without compromising global context, yielding an average improvement of 6.4\% on long-video benchmarks.

Faizan Farooq Khan, K J Joseph, Koustava Goswami et al.

Design generation, in its essence, is a step-by-step process where designers progressively refine and enhance their work through careful modifications. Despite this fundamental characteristic, existing approaches mainly treat design synthesis as a single-step generation problem, significantly underestimating the inherent complexity of the creative process. To bridge this gap, we propose a novel problem setting called Step-by-Step Layered Design Generation, which tasks a machine learning model with generating a design that adheres to a sequence of instructions from a designer. Leveraging recent advancements in multi-modal LLMs, we propose SLEDGE: Step-by-step LayEred Design GEnerator to model each update to a design as an atomic, layered change over its previous state, while being grounded in the instruction. To complement our new problem setting, we introduce a new evaluation suite, including a dataset and a benchmark. Our exhaustive experimental analysis and comparison with state-of-the-art approaches tailored to our new setup demonstrate the efficacy of our approach. We hope our work will attract attention to this pragmatic and under-explored research area.

Wenxuan Zhang, Lemeng Wu, Changsheng Zhao et al.

Diffusion Large Language Models (dLLMs) introduce a new paradigm for language generation, which in turn presents new challenges for aligning them with human preferences. In this work, we aim to improve the policy optimization for dLLMs by reducing the cost of the trajectory probability calculation, thereby enabling scaled-up offline policy training. We prove that: (i) under reference policy regularization, the probability ratio of the newly unmasked tokens is an unbiased estimate of that of intermediate diffusion states, and (ii) the probability of the full trajectory can be effectively estimated with a single forward pass of a re-masked final state. By integrating these two trajectory reduction strategies into a policy optimization objective, we propose Trajectory Reduction Policy Optimization (dTRPO). We evaluate dTRPO on 7B dLLMs across instruction-following and reasoning benchmarks. Results show that it substantially improves the core performance of state-of-the-art dLLMs, achieving gains of up to 9.6% on STEM tasks, up to 4.3% on coding tasks, and up to 3.0% on instruction-following tasks. Moreover, dTRPO exhibits strong training efficiency due to its offline, single-forward nature, and achieves improved generation efficiency through high-quality outputs.

Habib Slim, Shariq Farooq Bhat, Mohamed Elhoseiny et al.

Creating and editing high-quality 3D content remains a central challenge in computer graphics. We address this challenge by introducing CompoSE, a novel method for Compositional Synthesis and Editing of 3D shapes via part-aware control. Our method takes as input a set of coarse geometric primitives (e.g., bounding boxes) that represent distinct object parts arranged in a particular spatial configuration, and synthesizes as output part-separated 3D objects that support localized granular (i.e., compositional) editing of individual parts. The key insight that enables our method is our use of a diffusion transformer architecture that alternates between processing each part locally and aggregating contextual information across parts globally, and features a novel conditioning technique that ensures strong adherence to the user's input. Importantly, our method learns to infer part semantics and symmetries directly from the user's coarse layout guidance, and does not require part-level text prompts. We demonstrate that our method enables powerful part-level editing capabilities, including context-aware substitution, addition, deletion, and style-preserving resizing operations. We show through extensive experiments that our method significantly outperforms existing approaches on guided synthesis, as measured by objective metrics and LLM-based evaluations.

Xiaoqian Shen, Mohamed Elhoseiny

Recent generative models have demonstrated impressive capabilities in generating realistic and visually pleasing images grounded on textual prompts. Nevertheless, a significant challenge remains in applying these models for the more intricate task of story visualization. Since it requires resolving pronouns (he, she, they) in the frame descriptions, i.e., anaphora resolution, and ensuring consistent characters and background synthesis across frames. Yet, the emerging Large Language Model (LLM) showcases robust reasoning abilities to navigate through ambiguous references and process extensive sequences. Therefore, we introduce \emph{StoryGPT-V}, which leverages the merits of the latent diffusion (LDM) and LLM to produce images with consistent and high-quality characters grounded on given story descriptions. First, we train a character-aware LDM, which takes character-augmented semantic embedding as input and includes the supervision of the cross-attention map using character segmentation masks, aiming to enhance character generation accuracy and faithfulness. In the second stage, we enable an alignment between the output of LLM and the character-augmented embedding residing in the input space of the first-stage model. This harnesses the reasoning ability of LLM to address ambiguous references and the comprehension capability to memorize the context. We conduct comprehensive experiments on two visual story visualization benchmarks. Our model reports superior quantitative results and consistently generates accurate characters of remarkable quality with low memory consumption. Our code is publicly available at: \href{https://xiaoqian-shen.github.io/StoryGPT-V}{https://xiaoqian-shen.github.io/StoryGPT-V}.

Sanjoy Chowdhury, Sayan Nag, Subhrajyoti Dasgupta et al.

With the rapid advancement of Multi-modal Large Language Models (MLLMs), several diagnostic benchmarks have recently been developed to assess these models' multi-modal reasoning proficiency. However, these benchmarks are restricted to assessing primarily the visual aspect and do not examine the holistic audio-visual (AV) understanding. Moreover, currently, there are no benchmarks that investigate the capabilities of AVLLMs to calibrate their responses when presented with perturbed inputs. To this end, we introduce Audio-Visual Trustworthiness assessment Benchmark (AVTrustBench), comprising 600K samples spanning over 9 meticulously crafted tasks, evaluating the capabilities of AVLLMs across three distinct dimensions: Adversarial attack, Compositional reasoning, and Modality-specific dependency. Using our benchmark we extensively evaluate 13 state-of-the-art AVLLMs. The findings reveal that the majority of existing models fall significantly short of achieving human-like comprehension, offering valuable insights for future research directions. To alleviate the limitations in the existing approaches, we further propose a robust, model-agnostic calibrated audio-visual preference optimization based training strategy CAVPref, obtaining a gain up to 30.19% across all 9 tasks. We will publicly release our code and benchmark to facilitate future research in this direction.

Faizan Farooq Khan, Diana Kim, Divyansh Jha et al.

Discovering the creative potentials of a random signal to various artistic expressions in aesthetic and conceptual richness is a ground for the recent success of generative machine learning as a way of art creation. To understand the new artistic medium better, we conduct a comprehensive analysis to position AI-generated art within the context of human art heritage. Our comparative analysis is based on an extensive dataset, dubbed ``ArtConstellation,'' consisting of annotations about art principles, likability, and emotions for 6,000 WikiArt and 3,200 AI-generated artworks. After training various state-of-the-art generative models, art samples are produced and compared with WikiArt data on the last hidden layer of a deep-CNN trained for style classification. We actively examined the various art principles to interpret the neural representations and used them to drive the comparative knowledge about human and AI-generated art. A key finding in the semantic analysis is that AI-generated artworks are visually related to the principle concepts for modern period art made in 1800-2000. In addition, through Out-Of-Distribution (OOD) and In-Distribution (ID) detection in CLIP space, we find that AI-generated artworks are ID to human art when they depict landscapes and geometric abstract figures, while detected as OOD when the machine art consists of deformed and twisted figures. We observe that machine-generated art is uniquely characterized by incomplete and reduced figuration. Lastly, we conducted a human survey about emotional experience. Color composition and familiar subjects are the key factors of likability and emotions in art appreciation. We propose our whole methodologies and collected dataset as our analytical framework to contrast human and AI-generated art, which we refer to as ``ArtNeuralConstellation''. Code is available at: https://github.com/faixan-khan/ArtNeuralConstellation

Sanjoy Chowdhury, Hanan Gani, Nishit Anand et al.

Recent advancements in reasoning optimization have greatly enhanced the performance of large language models (LLMs). However, existing work fails to address the complexities of audio-visual scenarios, underscoring the need for further research. In this paper, we introduce AURELIA, a novel actor-critic based audio-visual (AV) reasoning framework that distills structured, step-by-step reasoning into AVLLMs at test time, improving their ability to process complex multi-modal inputs without additional training or fine-tuning. To further advance AVLLM reasoning skills, we present AVReasonBench, a challenging benchmark comprising 4500 audio-visual questions, each paired with detailed step-by-step reasoning. Our benchmark spans six distinct tasks, including AV-GeoIQ, which evaluates AV reasoning combined with geographical and cultural knowledge. Evaluating 18 AVLLMs on AVReasonBench reveals significant limitations in their multi-modal reasoning capabilities. Using AURELIA, we achieve up to a 100% relative improvement, demonstrating its effectiveness. This performance gain highlights the potential of reasoning-enhanced data generation for advancing AVLLMs in real-world applications. Our code and data will be publicly released at: https: //github.com/schowdhury671/aurelia.

Jun Chen, Dannong Xu, Junjie Fei et al.

Large multimodal models (LMMs) have achieved impressive progress in vision-language understanding, yet they face limitations in real-world applications requiring complex reasoning over a large number of images. Existing benchmarks for multi-image question-answering are limited in scope, each question is paired with only up to 30 images, which does not fully capture the demands of large-scale retrieval tasks encountered in the real-world usages. To reduce these gaps, we introduce two document haystack benchmarks, dubbed DocHaystack and InfoHaystack, designed to evaluate LMM performance on large-scale visual document retrieval and understanding. Additionally, we propose V-RAG, a novel, vision-centric retrieval-augmented generation (RAG) framework that leverages a suite of multimodal vision encoders, each optimized for specific strengths, and a dedicated question-document relevance module. V-RAG sets a new standard, with a 9% and 11% improvement in Recall@1 on the challenging DocHaystack-1000 and InfoHaystack-1000 benchmarks, respectively, compared to the previous best baseline models. Additionally, integrating V-RAG with LMMs enables them to efficiently operate across thousands of images, yielding significant improvements on our DocHaystack and InfoHaystack benchmarks. Our code and datasets are available at https://github.com/Vision-CAIR/dochaystacks