Search

Adversarial input image perturbation attacks have emerged as a significant threat to machine learning algorithms, particularly in image classification setting. These attacks involve subtle perturbations to input images that cause neural networks to misclassify the input images, even though the images remain easily recognizable to humans. One critical area where adversarial attacks have been demonstrated is in automotive systems where traffic sign classification and recognition is critical, and where misclassified images can cause autonomous systems to take wrong actions. This work presents a new class of adversarial attacks. Unlike existing work that has focused on adversarial perturbations that leverage human-made artifacts to cause the perturbations, such as adding stickers, paint, or shining flashlights at traffic signs, this work leverages nature-made artifacts: tree leaves. By leveraging nature-made artifacts, the new class of attacks has plausible deniability: a fall leaf stuck to a street sign could come from a near-by tree, rather than be placed there by an malicious human attacker. To evaluate the new class of the adversarial input image perturbation attacks, this work analyses how fall leaves can cause misclassification in street signs. The work evaluates various leaves from different species of trees, and considers various parameters such as size, color due to tree leaf type, and rotation. The work demonstrates high success rate for misclassification. The work also explores the correlation between successful attacks and how they affect the edge detection, which is critical in many image classification algorithms.

Deep learning-based automated contouring and treatment planning has been proven to improve the efficiency and accuracy of radiotherapy. However, conventional radiotherapy treatment planning process has the automated contouring and treatment planning as separate tasks. Moreover in deep learning (DL), the contouring and dose prediction tasks for automated treatment planning are done independently. In this study, we applied the multi-task learning (MTL) approach in order to seamlessly integrate automated contouring and voxel-based dose prediction tasks, as MTL can leverage common information between the two tasks and be able able to increase the efficiency of the automated tasks. We developed our MTL framework using the two datasets: in-house prostate cancer dataset and the publicly available head and neck cancer dataset, OpenKBP. Compared to the sequential DL contouring and treatment planning tasks, our proposed method using MTL improved the mean absolute difference of dose volume histogram metrics of prostate and head and neck sites by 19.82% and 16.33%, respectively. Our MTL model for automated contouring and dose prediction tasks demonstrated enhanced dose prediction performance while maintaining or sometimes even improving the contouring accuracy. Compared to the baseline automated contouring model with the dice score coefficients of 0.818 for prostate and 0.674 for head and neck datasets, our MTL approach achieved average scores of 0.824 and 0.716 for these datasets, respectively. Our study highlights the potential of the proposed automated contouring and planning using MTL to support the development of efficient and accurate automated treatment planning for radiotherapy.

Graphic visual content helps in promoting information communication and inspiration divergence. However, the interpretation of visual content currently relies mainly on humans' personal knowledge background, thereby affecting the quality and efficiency of information acquisition and understanding. To improve the quality and efficiency of visual information transmission and avoid the limitation of the observer due to the information cocoon, we propose CoVis, a collaborative framework for fine-grained visual understanding. By designing and implementing a cascaded dual-layer segmentation network coupled with a large-language-model (LLM) based content generator, the framework extracts as much knowledge as possible from an image. Then, it generates visual analytics for images, assisting observers in comprehending imagery from a more holistic perspective. Quantitative experiments and qualitative experiments based on 32 human participants indicate that the CoVis has better performance than current methods in feature extraction and can generate more comprehensive and detailed visual descriptions than current general-purpose large models.

In this work, we address a challenge in video inpainting: reconstructing occluded regions in dynamic, real-world scenarios. Motivated by the need for continuous human motion monitoring in healthcare settings, where facial features are frequently obscured, we propose a diffusion-based video-level inpainting model, DiffMVR. Our approach introduces a dynamic dual-guided image prompting system, leveraging adaptive reference frames to guide the inpainting process. This enables the model to capture both fine-grained details and smooth transitions between video frames, offering precise control over inpainting direction and significantly improving restoration accuracy in challenging, dynamic environments. DiffMVR represents a significant advancement in the field of diffusion-based inpainting, with practical implications for real-time applications in various dynamic settings.

Model merging has gained increasing attention as an efficient and effective technique for integrating task-specific weights from various tasks into a unified multi-task model without retraining or additional data. As a representative approach, Task Arithmetic (TA) has demonstrated that combining task vectors through arithmetic operations facilitates efficient capability transfer between different tasks. In this framework, task vectors are obtained by subtracting the parameter values of a pre-trained model from those of individually fine-tuned models initialized from it. Despite the notable effectiveness of TA, interference among task vectors can adversely affect the performance of the merged model. In this paper, we relax the constraints of Task Arithmetic Property and propose Task Consistency Property, which can be regarded as being free from task interference. Through theoretical derivation, we show that such a property can be approximately achieved by seeking orthogonal task vectors. Guiding by this insight, we propose Adaptive Weight Disentanglement (AWD), which decomposes traditional task vectors into a redundant vector and several disentangled task vectors. The primary optimization objective of AWD is to achieve orthogonality among the disentangled task vectors, thereby closely approximating the desired solution. Notably, these disentangled task vectors can be seamlessly integrated into existing merging methodologies. Experimental results demonstrate that our AWD consistently and significantly improves upon previous merging approaches, achieving state-of-the-art results. Our code is available at \href{https://github.com/FarisXiong/AWD.git}{https://github.com/FarisXiong/AWD.git}.

Visual communication, dating back to prehistoric cave paintings, is the use of visual elements to convey ideas and information. In today's visually saturated world, effective design demands an understanding of graphic design principles, visual storytelling, human psychology, and the ability to distill complex information into clear visuals. This dissertation explores how recent advancements in vision-language models (VLMs) can be leveraged to automate the creation of effective visual communication designs. Although generative models have made great progress in generating images from text, they still struggle to simplify complex ideas into clear, abstract visuals and are constrained by pixel-based outputs, which lack flexibility for many design tasks. To address these challenges, we constrain the models' operational space and introduce task-specific regularizations. We explore various aspects of visual communication, namely, sketches and visual abstraction, typography, animation, and visual inspiration., Comment: PhD Thesis

Supervised deep learning requires massive labeled datasets, but obtaining annotations is not always easy or possible, especially for dense tasks like semantic segmentation. To overcome this issue, numerous works explore Unsupervised Domain Adaptation (UDA), which uses a labeled dataset from another domain (source), or Semi-Supervised Learning (SSL), which trains on a partially labeled set. Despite the success of UDA and SSL, reaching supervised performance at a low annotation cost remains a notoriously elusive goal. To address this, we study the promising setting of Semi-Supervised Domain Adaptation (SSDA). We propose a simple SSDA framework that combines consistency regularization, pixel contrastive learning, and self-training to effectively utilize a few target-domain labels. Our method outperforms prior art in the popular GTA-to-Cityscapes benchmark and shows that as little as 50 target labels can suffice to achieve near-supervised performance. Additional results on Synthia-to-Cityscapes, GTA-to-BDD and Synthia-to-BDD further demonstrate the effectiveness and practical utility of the method. Lastly, we find that existing UDA and SSL methods are not well-suited for the SSDA setting and discuss design patterns to adapt them., Comment: 28 pages, 6 figures

Despite their promise to perform complex reasoning, large language models (LLMs) have been shown to have limited effectiveness in end-to-end planning. This has inspired an intriguing question: if these models cannot plan well, can they still contribute to the planning framework as a helpful plan evaluator? In this work, we generalize this question to consider LLMs augmented with visual understanding, i.e., Vision-Language Models (VLMs). We introduce PathEval, a novel benchmark evaluating VLMs as plan evaluators in complex path-planning scenarios. Succeeding in the benchmark requires a VLM to be able to abstract traits of optimal paths from the scenario description, demonstrate precise low-level perception on each path, and integrate this information to decide the better path. Our analysis of state-of-the-art VLMs reveals that these models face significant challenges on the benchmark. We observe that the VLMs can precisely abstract given scenarios to identify the desired traits and exhibit mixed performance in integrating the provided information. Yet, their vision component presents a critical bottleneck, with models struggling to perceive low-level details about a path. Our experimental results show that this issue cannot be trivially addressed via end-to-end fine-tuning; rather, task-specific discriminative adaptation of these vision encoders is needed for these VLMs to become effective path evaluators.

We present a simple template for designing generative diffusion model algorithms based on an interpretation of diffusion sampling as a sequence of random walks. Score-based diffusion models are widely used to generate high-quality images. Diffusion models have also been shown to yield state-of-the-art performance in many inverse problems. While these algorithms are often surprisingly simple, the theory behind them is not, and multiple complex theoretical justifications exist in the literature. Here, we provide a simple and largely self-contained theoretical justification for score-based-diffusion models that avoids using the theory of Markov chains or reverse diffusion, instead centering the theory of random walks and Tweedie's formula. This approach leads to unified algorithmic templates for network training and sampling. In particular, these templates cleanly separate training from sampling, e.g., the noise schedule used during training need not match the one used during sampling. We show that several existing diffusion models correspond to particular choices within this template and demonstrate that other, more straightforward algorithmic choices lead to effective diffusion models. The proposed framework has the added benefit of enabling conditional sampling without any likelihood approximation.

Match-cuts are powerful cinematic tools that create seamless transitions between scenes, delivering strong visual and metaphorical connections. However, crafting match-cuts is a challenging, resource-intensive process requiring deliberate artistic planning. In MatchDiffusion, we present the first training-free method for match-cut generation using text-to-video diffusion models. MatchDiffusion leverages a key property of diffusion models: early denoising steps define the scene's broad structure, while later steps add details. Guided by this insight, MatchDiffusion employs "Joint Diffusion" to initialize generation for two prompts from shared noise, aligning structure and motion. It then applies "Disjoint Diffusion", allowing the videos to diverge and introduce unique details. This approach produces visually coherent videos suited for match-cuts. User studies and metrics demonstrate MatchDiffusion's effectiveness and potential to democratize match-cut creation., Comment: https://matchdiffusion.github.io

We introduce GaussianSpeech, a novel approach that synthesizes high-fidelity animation sequences of photo-realistic, personalized 3D human head avatars from spoken audio. To capture the expressive, detailed nature of human heads, including skin furrowing and finer-scale facial movements, we propose to couple speech signal with 3D Gaussian splatting to create realistic, temporally coherent motion sequences. We propose a compact and efficient 3DGS-based avatar representation that generates expression-dependent color and leverages wrinkle- and perceptually-based losses to synthesize facial details, including wrinkles that occur with different expressions. To enable sequence modeling of 3D Gaussian splats with audio, we devise an audio-conditioned transformer model capable of extracting lip and expression features directly from audio input. Due to the absence of high-quality datasets of talking humans in correspondence with audio, we captured a new large-scale multi-view dataset of audio-visual sequences of talking humans with native English accents and diverse facial geometry. GaussianSpeech consistently achieves state-of-the-art performance with visually natural motion at real time rendering rates, while encompassing diverse facial expressions and styles., Comment: Paper Video: https://youtu.be/2VqYoFlYcwQ Project Page: https://shivangi-aneja.github.io/projects/gaussianspeech

Numerous works have recently integrated 3D camera control into foundational text-to-video models, but the resulting camera control is often imprecise, and video generation quality suffers. In this work, we analyze camera motion from a first principles perspective, uncovering insights that enable precise 3D camera manipulation without compromising synthesis quality. First, we determine that motion induced by camera movements in videos is low-frequency in nature. This motivates us to adjust train and test pose conditioning schedules, accelerating training convergence while improving visual and motion quality. Then, by probing the representations of an unconditional video diffusion transformer, we observe that they implicitly perform camera pose estimation under the hood, and only a sub-portion of their layers contain the camera information. This suggested us to limit the injection of camera conditioning to a subset of the architecture to prevent interference with other video features, leading to 4x reduction of training parameters, improved training speed and 10% higher visual quality. Finally, we complement the typical dataset for camera control learning with a curated dataset of 20K diverse dynamic videos with stationary cameras. This helps the model disambiguate the difference between camera and scene motion, and improves the dynamics of generated pose-conditioned videos. We compound these findings to design the Advanced 3D Camera Control (AC3D) architecture, the new state-of-the-art model for generative video modeling with camera control., Comment: Project Page: https://snap-research.github.io/ac3d/

Knowledge distillation (KD) is the de facto standard for compressing large-scale models into smaller ones. Prior works have explored ever more complex KD strategies involving different objective functions, teacher-ensembles, and weight inheritance. In this work we explore an alternative, yet simple approach -- active data curation as effective distillation for contrastive multimodal pretraining. Our simple online batch selection method, ACID, outperforms strong KD baselines across various model-, data- and compute-configurations. Further, we find such an active data curation strategy to in fact be complementary to standard KD, and can be effectively combined to train highly performant inference-efficient models. Our simple and scalable pretraining framework, ACED, achieves state-of-the-art results across 27 zero-shot classification and retrieval tasks with upto 11% less inference FLOPs. We further demonstrate that our ACED models yield strong vision-encoders for training generative multimodal models in the LiT-Decoder setting, outperforming larger vision encoders for image-captioning and visual question-answering tasks.

Medical vision-language model models often struggle with generating accurate quantitative measurements in radiology reports, leading to hallucinations that undermine clinical reliability. We introduce FactCheXcker, a modular framework that de-hallucinates radiology report measurements by leveraging an improved query-code-update paradigm. Specifically, FactCheXcker employs specialized modules and the code generation capabilities of large language models to solve measurement queries generated based on the original report. After extracting measurable findings, the results are incorporated into an updated report. We evaluate FactCheXcker on endotracheal tube placement, which accounts for an average of 78% of report measurements, using the MIMIC-CXR dataset and 11 medical report-generation models. Our results show that FactCheXcker significantly reduces hallucinations, improves measurement precision, and maintains the quality of the original reports. Specifically, FactCheXcker improves the performance of all 11 models and achieves an average improvement of 94.0% in reducing measurement hallucinations measured by mean absolute error.

In this paper, we present TAPTRv3, which is built upon TAPTRv2 to improve its point tracking robustness in long videos. TAPTRv2 is a simple DETR-like framework that can accurately track any point in real-world videos without requiring cost-volume. TAPTRv3 improves TAPTRv2 by addressing its shortage in querying high quality features from long videos, where the target tracking points normally undergo increasing variation over time. In TAPTRv3, we propose to utilize both spatial and temporal context to bring better feature querying along the spatial and temporal dimensions for more robust tracking in long videos. For better spatial feature querying, we present Context-aware Cross-Attention (CCA), which leverages surrounding spatial context to enhance the quality of attention scores when querying image features. For better temporal feature querying, we introduce Visibility-aware Long-Temporal Attention (VLTA) to conduct temporal attention to all past frames while considering their corresponding visibilities, which effectively addresses the feature drifting problem in TAPTRv2 brought by its RNN-like long-temporal modeling. TAPTRv3 surpasses TAPTRv2 by a large margin on most of the challenging datasets and obtains state-of-the-art performance. Even when compared with methods trained with large-scale extra internal data, TAPTRv3 is still competitive.

Foundation models like ChatGPT and Sora that are trained on a huge scale of data have made a revolutionary social impact. However, it is extremely challenging for sensors in many different fields to collect similar scales of natural images to train strong foundation models. To this end, this work presents a simple and effective framework, SimCMF, to study an important problem: cross-modal fine-tuning from vision foundation models trained on natural RGB images to other imaging modalities of different physical properties (e.g., polarization). In SimCMF, we conduct a thorough analysis of different basic components from the most naive design and ultimately propose a novel cross-modal alignment module to address the modality misalignment problem. We apply SimCMF to a representative vision foundation model Segment Anything Model (SAM) to support any evaluated new imaging modality. Given the absence of relevant benchmarks, we construct a benchmark for performance evaluation. Our experiments confirm the intriguing potential of transferring vision foundation models in enhancing other sensors' performance. SimCMF can improve the segmentation performance (mIoU) from 22.15% to 53.88% on average for evaluated modalities and consistently outperforms other baselines. The code is available at https://github.com/mt-cly/SimCMF, Comment: project page: https://mt-cly.github.io/SimCMF.github.io/. arXiv admin note: substantial text overlap with arXiv:2409.08083

Generating long-duration videos has always been a significant challenge due to the inherent complexity of spatio-temporal domain and the substantial GPU memory demands required to calculate huge size tensors. While diffusion based generative models achieve state-of-the-art performance in video generation task, they are typically trained with predefined video resolutions and lengths. During inference, a noise tensor with specific resolution and length should be specified at first, and the model will perform denoising on the entire video tensor simultaneously, all the frames together. Such approach will easily raise an out-of-memory (OOM) problem when the specified resolution and/or length exceed a certain limit. One of the solutions to this problem is to generate many short video chunks autoregressively with strong inter-chunk spatio-temporal relation and then concatenate them together to form a long video. In this approach, a long video generation task is divided into multiple short video generation subtasks, and the cost of each subtask is reduced to a feasible level. In this paper, we conduct a detailed survey on long video generation with the autoregressive chunk-by-chunk strategy. We address common problems caused by applying short image-to-video models to long video tasks and design an efficient $k$-step search solution to mitigate these problems.

Pre-training on large-scale unlabeled datasets contribute to the model achieving powerful performance on 3D vision tasks, especially when annotations are limited. However, existing rendering-based self-supervised frameworks are computationally demanding and memory-intensive during pre-training due to the inherent nature of volume rendering. In this paper, we propose an efficient framework named GS$^3$ to learn point cloud representation, which seamlessly integrates fast 3D Gaussian Splatting into the rendering-based framework. The core idea behind our framework is to pre-train the point cloud encoder by comparing rendered RGB images with real RGB images, as only Gaussian points enriched with learned rich geometric and appearance information can produce high-quality renderings. Specifically, we back-project the input RGB-D images into 3D space and use a point cloud encoder to extract point-wise features. Then, we predict 3D Gaussian points of the scene from the learned point cloud features and uses a tile-based rasterizer for image rendering. Finally, the pre-trained point cloud encoder can be fine-tuned to adapt to various downstream 3D tasks, including high-level perception tasks such as 3D segmentation and detection, as well as low-level tasks such as 3D scene reconstruction. Extensive experiments on downstream tasks demonstrate the strong transferability of the pre-trained point cloud encoder and the effectiveness of our self-supervised learning framework. In addition, our GS$^3$ framework is highly efficient, achieving approximately 9$\times$ pre-training speedup and less than 0.25$\times$ memory cost compared to the previous rendering-based framework Ponder., Comment: 14 pages, 4 figures, 15 tables

Recent work has shown that diffusion models can be used as powerful neural rendering engines that can be leveraged for inserting virtual objects into images. Unlike typical physics-based renderers, however, neural rendering engines are limited by the lack of manual control over the lighting setup, which is often essential for improving or personalizing the desired image outcome. In this paper, we show that precise lighting control can be achieved for object relighting simply by specifying the desired shadows of the object. Rather surprisingly, we show that injecting only the shadow of the object into a pre-trained diffusion-based neural renderer enables it to accurately shade the object according to the desired light position, while properly harmonizing the object (and its shadow) within the target background image. Our method, SpotLight, leverages existing neural rendering approaches and achieves controllable relighting results with no additional training. Specifically, we demonstrate its use with two neural renderers from the recent literature. We show that SpotLight achieves superior object compositing results, both quantitatively and perceptually, as confirmed by a user study, outperforming existing diffusion-based models specifically designed for relighting., Comment: Project page: https://lvsn.github.io/spotlight

3D vision-language (VL) reasoning has gained significant attention due to its potential to bridge the 3D physical world with natural language descriptions. Existing approaches typically follow task-specific, highly specialized paradigms. Therefore, these methods focus on a limited range of reasoning sub-tasks and rely heavily on the hand-crafted modules and auxiliary losses. This highlights the need for a simpler, unified and general-purpose model. In this paper, we leverage the inherent connection between 3D scene graphs and natural language, proposing a 3D scene graph-guided vision-language pre-training (VLP) framework. Our approach utilizes modality encoders, graph convolutional layers and cross-attention layers to learn universal representations that adapt to a variety of 3D VL reasoning tasks, thereby eliminating the need for task-specific designs. The pre-training objectives include: 1) Scene graph-guided contrastive learning, which leverages the strong correlation between 3D scene graphs and natural language to align 3D objects with textual features at various fine-grained levels; and 2) Masked modality learning, which uses cross-modality information to reconstruct masked words and 3D objects. Instead of directly reconstructing the 3D point clouds of masked objects, we use position clues to predict their semantic categories. Extensive experiments demonstrate that our pre-training model, when fine-tuned on several downstream tasks, achieves performance comparable to or better than existing methods in tasks such as 3D visual grounding, 3D dense captioning, and 3D question answering., Comment: 14 pages, 8 figures, 7 tables

Generating realistic 3D human-object interactions (HOIs) from text descriptions is a active research topic with potential applications in virtual and augmented reality, robotics, and animation. However, creating high-quality 3D HOIs remains challenging due to the lack of large-scale interaction data and the difficulty of ensuring physical plausibility, especially in out-of-domain (OOD) scenarios. Current methods tend to focus either on the body or the hands, which limits their ability to produce cohesive and realistic interactions. In this paper, we propose OOD-HOI, a text-driven framework for generating whole-body human-object interactions that generalize well to new objects and actions. Our approach integrates a dual-branch reciprocal diffusion model to synthesize initial interaction poses, a contact-guided interaction refiner to improve physical accuracy based on predicted contact areas, and a dynamic adaptation mechanism which includes semantic adjustment and geometry deformation to improve robustness. Experimental results demonstrate that our OOD-HOI could generate more realistic and physically plausible 3D interaction pose in OOD scenarios compared to existing methods.

Diffusion models have emerged as a powerful tool for generating high-quality images, videos, and 3D content. While sampling guidance techniques like CFG improve quality, they reduce diversity and motion. Autoguidance mitigates these issues but demands extra weak model training, limiting its practicality for large-scale models. In this work, we introduce Spatiotemporal Skip Guidance (STG), a simple training-free sampling guidance method for enhancing transformer-based video diffusion models. STG employs an implicit weak model via self-perturbation, avoiding the need for external models or additional training. By selectively skipping spatiotemporal layers, STG produces an aligned, degraded version of the original model to boost sample quality without compromising diversity or dynamic degree. Our contributions include: (1) introducing STG as an efficient, high-performing guidance technique for video diffusion models, (2) eliminating the need for auxiliary models by simulating a weak model through layer skipping, and (3) ensuring quality-enhanced guidance without compromising sample diversity or dynamics unlike CFG. For additional results, visit https://junhahyung.github.io/STGuidance., Comment: project page: https://junhahyung.github.io/STGuidance

Combining the complementary benefits of frames and events has been widely used for object detection in challenging scenarios. However, most object detection methods use two independent Artificial Neural Network (ANN) branches, limiting cross-modality information interaction across the two visual streams and encountering challenges in extracting temporal cues from event streams with low power consumption. To address these challenges, we propose HDI-Former, a Hybrid Dynamic Interaction ANN-SNN Transformer, marking the first trial to design a directly trained hybrid ANN-SNN architecture for high-accuracy and energy-efficient object detection using frames and events. Technically, we first present a novel semantic-enhanced self-attention mechanism that strengthens the correlation between image encoding tokens within the ANN Transformer branch for better performance. Then, we design a Spiking Swin Transformer branch to model temporal cues from event streams with low power consumption. Finally, we propose a bio-inspired dynamic interaction mechanism between ANN and SNN sub-networks for cross-modality information interaction. The results demonstrate that our HDI-Former outperforms eleven state-of-the-art methods and our four baselines by a large margin. Our SNN branch also shows comparable performance to the ANN with the same architecture while consuming 10.57$\times$ less energy on the DSEC-Detection dataset. Our open-source code is available in the supplementary material., Comment: 17 pages, 11 figures

Text-to-image diffusion models have emerged as powerful priors for real-world image super-resolution (Real-ISR). However, existing methods may produce unintended results due to noisy text prompts and their lack of spatial information. In this paper, we present HoliSDiP, a framework that leverages semantic segmentation to provide both precise textual and spatial guidance for diffusion-based Real-ISR. Our method employs semantic labels as concise text prompts while introducing dense semantic guidance through segmentation masks and our proposed Segmentation-CLIP Map. Extensive experiments demonstrate that HoliSDiP achieves significant improvement in image quality across various Real-ISR scenarios through reduced prompt noise and enhanced spatial control., Comment: Project page: https://liyuantsao.github.io/HoliSDiP/

Large vision-language models (LVLMs) have demonstrated exceptional performance on complex multimodal tasks. However, they continue to suffer from significant hallucination issues, including object, attribute, and relational hallucinations. To accurately detect these hallucinations, we investigated the variations in cross-modal attention patterns between hallucination and non-hallucination states. Leveraging these distinctions, we developed a lightweight detector capable of identifying hallucinations. Our proposed method, Detecting Hallucinations by Cross-modal Attention Patterns (DHCP), is straightforward and does not require additional LVLM training or extra LVLM inference steps. Experimental results show that DHCP achieves remarkable performance in hallucination detection. By offering novel insights into the identification and analysis of hallucinations in LVLMs, DHCP contributes to advancing the reliability and trustworthiness of these models., Comment: 18 pages, 5 figures

Recently, text-to-motion models have opened new possibilities for creating realistic human motion with greater efficiency and flexibility. However, aligning motion generation with event-level textual descriptions presents unique challenges due to the complex relationship between textual prompts and desired motion outcomes. To address this, we introduce AToM, a framework that enhances the alignment between generated motion and text prompts by leveraging reward from GPT-4Vision. AToM comprises three main stages: Firstly, we construct a dataset MotionPrefer that pairs three types of event-level textual prompts with generated motions, which cover the integrity, temporal relationship and frequency of motion. Secondly, we design a paradigm that utilizes GPT-4Vision for detailed motion annotation, including visual data formatting, task-specific instructions and scoring rules for each sub-task. Finally, we fine-tune an existing text-to-motion model using reinforcement learning guided by this paradigm. Experimental results demonstrate that AToM significantly improves the event-level alignment quality of text-to-motion generation.

3D Gaussian Splatting (3DGS) has recently emerged as a state-of-the-art 3D reconstruction and rendering technique due to its high-quality results and fast training and rendering time. However, pixels covered by the same Gaussian are always shaded in the same color up to a Gaussian falloff scaling factor. Furthermore, the finest geometric detail any individual Gaussian can represent is a simple ellipsoid. These properties of 3DGS greatly limit the expressivity of individual Gaussian primitives. To address these issues, we draw inspiration from texture and alpha mapping in traditional graphics and integrate it with 3DGS. Specifically, we propose a new generalized Gaussian appearance representation that augments each Gaussian with alpha~(A), RGB, or RGBA texture maps to model spatially varying color and opacity across the extent of each Gaussian. As such, each Gaussian can represent a richer set of texture patterns and geometric structures, instead of just a single color and ellipsoid as in naive Gaussian Splatting. Surprisingly, we found that the expressivity of Gaussians can be greatly improved by using alpha-only texture maps, and further augmenting Gaussians with RGB texture maps achieves the highest expressivity. We validate our method on a wide variety of standard benchmark datasets and our own custom captures at both the object and scene levels. We demonstrate image quality improvements over existing methods while using a similar or lower number of Gaussians., Comment: Project website: https://textured-gaussians.github.io/

Given a single in-the-wild human photo, it remains a challenging task to reconstruct a high-fidelity 3D human model. Existing methods face difficulties including a) the varying body proportions captured by in-the-wild human images; b) diverse personal belongings within the shot; and c) ambiguities in human postures and inconsistency in human textures. In addition, the scarcity of high-quality human data intensifies the challenge. To address these problems, we propose a Generalizable image-to-3D huMAN reconstruction framework, dubbed GeneMAN, building upon a comprehensive multi-source collection of high-quality human data, including 3D scans, multi-view videos, single photos, and our generated synthetic human data. GeneMAN encompasses three key modules. 1) Without relying on parametric human models (e.g., SMPL), GeneMAN first trains a human-specific text-to-image diffusion model and a view-conditioned diffusion model, serving as GeneMAN 2D human prior and 3D human prior for reconstruction, respectively. 2) With the help of the pretrained human prior models, the Geometry Initialization-&-Sculpting pipeline is leveraged to recover high-quality 3D human geometry given a single image. 3) To achieve high-fidelity 3D human textures, GeneMAN employs the Multi-Space Texture Refinement pipeline, consecutively refining textures in the latent and the pixel spaces. Extensive experimental results demonstrate that GeneMAN could generate high-quality 3D human models from a single image input, outperforming prior state-of-the-art methods. Notably, GeneMAN could reveal much better generalizability in dealing with in-the-wild images, often yielding high-quality 3D human models in natural poses with common items, regardless of the body proportions in the input images., Comment: Project page: https://roooooz.github.io/GeneMAN/

3D geometric information is essential for manipulation tasks, as robots need to perceive the 3D environment, reason about spatial relationships, and interact with intricate spatial configurations. Recent research has increasingly focused on the explicit extraction of 3D features, while still facing challenges such as the lack of large-scale robotic 3D data and the potential loss of spatial geometry. To address these limitations, we propose the Lift3D framework, which progressively enhances 2D foundation models with implicit and explicit 3D robotic representations to construct a robust 3D manipulation policy. Specifically, we first design a task-aware masked autoencoder that masks task-relevant affordance patches and reconstructs depth information, enhancing the 2D foundation model's implicit 3D robotic representation. After self-supervised fine-tuning, we introduce a 2D model-lifting strategy that establishes a positional mapping between the input 3D points and the positional embeddings of the 2D model. Based on the mapping, Lift3D utilizes the 2D foundation model to directly encode point cloud data, leveraging large-scale pretrained knowledge to construct explicit 3D robotic representations while minimizing spatial information loss. In experiments, Lift3D consistently outperforms previous state-of-the-art methods across several simulation benchmarks and real-world scenarios.

In the 21st-century information age, with the development of big data technology, effectively extracting valuable information from massive data has become a key issue. Traditional data mining methods are inadequate when faced with large-scale, high-dimensional and complex data. Especially when labeled data is scarce, their performance is greatly limited. This study optimizes data mining algorithms by introducing semi-supervised learning methods, aiming to improve the algorithm's ability to utilize unlabeled data, thereby achieving more accurate data analysis and pattern recognition under limited labeled data conditions. Specifically, we adopt a self-training method and combine it with a convolutional neural network (CNN) for image feature extraction and classification, and continuously improve the model prediction performance through an iterative process. The experimental results demonstrate that the proposed method significantly outperforms traditional machine learning techniques such as Support Vector Machine (SVM), XGBoost, and Multi-Layer Perceptron (MLP) on the CIFAR-10 image classification dataset. Notable improvements were observed in key performance metrics, including accuracy, recall, and F1 score. Furthermore, the robustness and noise-resistance capabilities of the semi-supervised CNN model were validated through experiments under varying noise levels, confirming its practical applicability in real-world scenarios.

The recent advancements in auto-regressive multimodal large language models (MLLMs) have demonstrated promising progress for vision-language tasks. While there exists a variety of studies investigating the processing of linguistic information within large language models, little is currently known about the inner working mechanism of MLLMs and how linguistic and visual information interact within these models. In this study, we aim to fill this gap by examining the information flow between different modalities -- language and vision -- in MLLMs, focusing on visual question answering. Specifically, given an image-question pair as input, we investigate where in the model and how the visual and linguistic information are combined to generate the final prediction. Conducting experiments with a series of models from the LLaVA series, we find that there are two distinct stages in the process of integration of the two modalities. In the lower layers, the model first transfers the more general visual features of the whole image into the representations of (linguistic) question tokens. In the middle layers, it once again transfers visual information about specific objects relevant to the question to the respective token positions of the question. Finally, in the higher layers, the resulting multimodal representation is propagated to the last position of the input sequence for the final prediction. Overall, our findings provide a new and comprehensive perspective on the spatial and functional aspects of image and language processing in the MLLMs, thereby facilitating future research into multimodal information localization and editing.

Text-to-image diffusion models produce impressive results but are frustrating tools for artists who desire fine-grained control. For example, a common use case is to create images of a specific instance in novel contexts, i.e., "identity-preserving generation". This setting, along with many other tasks (e.g., relighting), is a natural fit for image+text-conditional generative models. However, there is insufficient high-quality paired data to train such a model directly. We propose Diffusion Self-Distillation, a method for using a pre-trained text-to-image model to generate its own dataset for text-conditioned image-to-image tasks. We first leverage a text-to-image diffusion model's in-context generation ability to create grids of images and curate a large paired dataset with the help of a Visual-Language Model. We then fine-tune the text-to-image model into a text+image-to-image model using the curated paired dataset. We demonstrate that Diffusion Self-Distillation outperforms existing zero-shot methods and is competitive with per-instance tuning techniques on a wide range of identity-preservation generation tasks, without requiring test-time optimization., Comment: Project page: https://primecai.github.io/dsd/

We present CAT4D, a method for creating 4D (dynamic 3D) scenes from monocular video. CAT4D leverages a multi-view video diffusion model trained on a diverse combination of datasets to enable novel view synthesis at any specified camera poses and timestamps. Combined with a novel sampling approach, this model can transform a single monocular video into a multi-view video, enabling robust 4D reconstruction via optimization of a deformable 3D Gaussian representation. We demonstrate competitive performance on novel view synthesis and dynamic scene reconstruction benchmarks, and highlight the creative capabilities for 4D scene generation from real or generated videos. See our project page for results and interactive demos: \url{cat-4d.github.io}., Comment: Project page: https://cat-4d.github.io/

Advancing towards generalist agents necessitates the concurrent processing of multiple tasks using a unified model, thereby underscoring the growing significance of simultaneous model training on multiple downstream tasks. A common issue in multi-task learning is the occurrence of gradient conflict, which leads to potential competition among different tasks during joint training. This competition often results in improvements in one task at the expense of deterioration in another. Although several optimization methods have been developed to address this issue by manipulating task gradients for better task balancing, they cannot decrease the incidence of gradient conflict. In this paper, we systematically investigate the occurrence of gradient conflict across different methods and propose a strategy to reduce such conflicts through sparse training (ST), wherein only a portion of the model's parameters are updated during training while keeping the rest unchanged. Our extensive experiments demonstrate that ST effectively mitigates conflicting gradients and leads to superior performance. Furthermore, ST can be easily integrated with gradient manipulation techniques, thus enhancing their effectiveness.

Purpose: To explore best-practice approaches for generating synthetic chest X-ray images and augmenting medical imaging datasets to optimize the performance of deep learning models in downstream tasks like classification and segmentation. Materials and Methods: We utilized a latent diffusion model to condition the generation of synthetic chest X-rays on text prompts and/or segmentation masks. We explored methods like using a proxy model and using radiologist feedback to improve the quality of synthetic data. These synthetic images were then generated from relevant disease information or geometrically transformed segmentation masks and added to ground truth training set images from the CheXpert, CANDID-PTX, SIIM, and RSNA Pneumonia datasets to measure improvements in classification and segmentation model performance on the test sets. F1 and Dice scores were used to evaluate classification and segmentation respectively. One-tailed t-tests with Bonferroni correction assessed the statistical significance of performance improvements with synthetic data. Results: Across all experiments, the synthetic data we generated resulted in a maximum mean classification F1 score improvement of 0.150453 (CI: 0.099108-0.201798; P=0.0031) compared to using only real data. For segmentation, the maximum Dice score improvement was 0.14575 (CI: 0.108267-0.183233; P=0.0064). Conclusion: Best practices for generating synthetic chest X-ray images for downstream tasks include conditioning on single-disease labels or geometrically transformed segmentation masks, as well as potentially using proxy modeling for fine-tuning such generations.

We introduce a structured light system that captures full-frame depth at rates of a thousand frames per second, four times faster than the previous state of the art. Our key innovation to this end is the design of an acousto-optic light scanning device that can scan light planes at rates up to two million planes per second. We combine this device with an event camera for structured light, using the sparse events triggered on the camera as we sweep a light plane on the scene for depth triangulation. In contrast to prior work, where light scanning is the bottleneck towards faster structured light operation, our light scanning device is three orders of magnitude faster than the event camera's full-frame bandwidth, thus allowing us to take full advantage of the event camera's fast operation. To surpass this bandwidth, we additionally demonstrate adaptive scanning of only regions of interest, at speeds an order of magnitude faster than the theoretical full-frame limit for event cameras.

The search for evidence of past life on Mars presents a tremendous challenge that requires the usage of very advanced robotic technologies to overcome it. Current digital microscopic imagers and spectrometers used for astrobiological examination suffer from limitations such as insufficient resolution, narrow detection range, and lack of portability. To overcome these challenges, this research study presents modifications to the Phoenix rover to expand its capability for detecting biosignatures on Mars. This paper examines the modifications implemented on the Phoenix rover to enhance its capability to detect a broader spectrum of biosignatures. One of the notable improvements comprises the integration of advanced digital microscopic imagers and spectrometers, enabling high-resolution examination of soil samples. Additionally, the mechanical components of the device have been reinforced to enhance maneuverability and optimize subsurface sampling capabilities. Empirical investigations have demonstrated that Phoenix has the capability to navigate diverse geological environments and procure samples for the purpose of biomolecular analysis. The biomolecular instrumentation and hybrid analytical methods showcased in this study demonstrate considerable potential for future astrobiology missions on Mars. The potential for enhancing the system lies in the possibility of broadening the range of detectable biomarkers and biosignatures., Comment: Key Words : Mars, Rover, Phoenix, Biosignatures, Biomolecular Analysis, Microscopy, Spectroscopy, Sampling, Astrobiology

While vision transformers show promise in numerous image restoration (IR) tasks, the challenge remains in efficiently generalizing and scaling up a model for multiple IR tasks. To strike a balance between efficiency and model capacity for a generalized transformer-based IR method, we propose a hierarchical information flow mechanism for image restoration, dubbed Hi-IR, which progressively propagates information among pixels in a bottom-up manner. Hi-IR constructs a hierarchical information tree representing the degraded image across three levels. Each level encapsulates different types of information, with higher levels encompassing broader objects and concepts and lower levels focusing on local details. Moreover, the hierarchical tree architecture removes long-range self-attention, improves the computational efficiency and memory utilization, thus preparing it for effective model scaling. Based on that, we explore model scaling to improve our method's capabilities, which is expected to positively impact IR in large-scale training settings. Extensive experimental results show that Hi-IR achieves state-of-the-art performance in seven common image restoration tasks, affirming its effectiveness and generalizability.

Face manipulation detection has been receiving a lot of attention for the reliability and security of the face images/videos. Recent studies focus on using auxiliary information or prior knowledge to capture robust manipulation traces, which are shown to be promising. As one of the important face features, the face depth map, which has shown to be effective in other areas such as face recognition or face detection, is unfortunately paid little attention to in literature for face manipulation detection. In this paper, we explore the possibility of incorporating the face depth map as auxiliary information for robust face manipulation detection. To this end, we first propose a Face Depth Map Transformer (FDMT) to estimate the face depth map patch by patch from an RGB face image, which is able to capture the local depth anomaly created due to manipulation. The estimated face depth map is then considered as auxiliary information to be integrated with the backbone features using a Multi-head Depth Attention (MDA) mechanism that is newly designed. We also propose an RGB-Depth Inconsistency Attention (RDIA) module to effectively capture the inter-frame inconsistency for multi-frame input. Various experiments demonstrate the advantage of our proposed method for face manipulation detection., Comment: 12 pages, 10 figures. arXiv admin note: substantial text overlap with arXiv:2212.14230

Dexterous manipulation with contact-rich interactions is crucial for advanced robotics. While recent diffusion-based planning approaches show promise for simpler manipulation tasks, they often produce unrealistic ghost states (e.g., the object automatically moves without hand contact) or lack adaptability when handling complex sequential interactions. In this work, we introduce DexDiffuser, an interaction-aware diffusion planning framework for adaptive dexterous manipulation. DexDiffuser models joint state-action dynamics through a dual-phase diffusion process which consists of pre-interaction contact alignment and post-contact goal-directed control, enabling goal-adaptive generalizable dexterous manipulation. Additionally, we incorporate dynamics model-based dual guidance and leverage large language models for automated guidance function generation, enhancing generalizability for physical interactions and facilitating diverse goal adaptation through language cues. Experiments on physical interaction tasks such as door opening, pen and block re-orientation, and hammer striking demonstrate DexDiffuser's effectiveness on goals outside training distributions, achieving over twice the average success rate (59.2% vs. 29.5%) compared to existing methods. Our framework achieves 70.0% success on 30-degree door opening, 40.0% and 36.7% on pen and block half-side re-orientation respectively, and 46.7% on hammer nail half drive, highlighting its robustness and flexibility in contact-rich manipulation., Comment: 27 pages. Project page: https://dexdiffuser.github.io/

Diffusion models are proficient at generating high-quality images. They are however effective only when operating at the resolution used during training. Inference at a scaled resolution leads to repetitive patterns and structural distortions. Retraining at higher resolutions quickly becomes prohibitive. Thus, methods enabling pre-existing diffusion models to operate at flexible test-time resolutions are highly desirable. Previous works suffer from frequent artifacts and often introduce large latency overheads. We propose two simple modules that combine to solve these issues. We introduce a Frequency Modulation (FM) module that leverages the Fourier domain to improve the global structure consistency, and an Attention Modulation (AM) module which improves the consistency of local texture patterns, a problem largely ignored in prior works. Our method, coined Fam diffusion, can seamlessly integrate into any latent diffusion model and requires no additional training. Extensive qualitative results highlight the effectiveness of our method in addressing structural and local artifacts, while quantitative results show state-of-the-art performance. Also, our method avoids redundant inference tricks for improved consistency such as patch-based or progressive generation, leading to negligible latency overheads.

We present PhyCAGE, the first approach for physically plausible compositional 3D asset generation from a single image. Given an input image, we first generate consistent multi-view images for components of the assets. These images are then fitted with 3D Gaussian Splatting representations. To ensure that the Gaussians representing objects are physically compatible with each other, we introduce a Physical Simulation-Enhanced Score Distillation Sampling (PSE-SDS) technique to further optimize the positions of the Gaussians. It is achieved by setting the gradient of the SDS loss as the initial velocity of the physical simulation, allowing the simulator to act as a physics-guided optimizer that progressively corrects the Gaussians' positions to a physically compatible state. Experimental results demonstrate that the proposed method can generate physically plausible compositional 3D assets given a single image., Comment: Project page: https://wolfball.github.io/phycage/

Visual Language Navigation is a task that challenges robots to navigate in realistic environments based on natural language instructions. While previous research has largely focused on static settings, real-world navigation must often contend with dynamic human obstacles. Hence, we propose an extension to the task, termed Adaptive Visual Language Navigation (AdaVLN), which seeks to narrow this gap. AdaVLN requires robots to navigate complex 3D indoor environments populated with dynamically moving human obstacles, adding a layer of complexity to navigation tasks that mimic the real-world. To support exploration of this task, we also present AdaVLN simulator and AdaR2R datasets. The AdaVLN simulator enables easy inclusion of fully animated human models directly into common datasets like Matterport3D. We also introduce a "freeze-time" mechanism for both the navigation task and simulator, which pauses world state updates during agent inference, enabling fair comparisons and experimental reproducibility across different hardware. We evaluate several baseline models on this task, analyze the unique challenges introduced by AdaVLN, and demonstrate its potential to bridge the sim-to-real gap in VLN research.

The patterns on wafer maps play a crucial role in helping engineers identify the causes of production issues during semiconductor manufacturing. In order to reduce costs and improve accuracy, automation technology is essential, and recent developments in deep learning have led to impressive results in wafer map pattern recognition. In this context, inspired by the effectiveness of semi-supervised learning and contrastive learning methods, we introduce an innovative approach that integrates the Mean Teacher framework with the supervised contrastive learning loss for enhanced wafer map pattern recognition. Our methodology not only addresses the nuances of wafer patterns but also tackles challenges arising from limited labeled data. To further refine the process, we address data imbalance in the wafer dataset by employing SMOTE and under-sampling techniques. We conduct a comprehensive analysis of our proposed method and demonstrate its effectiveness through experiments using real-world dataset WM811K obtained from semiconductor manufacturers. Compared to the baseline method, our method has achieved 5.46%, 6.68%, 5.42%, and 4.53% improvements in Accuracy, Precision, Recall, and F1 score, respectively., Comment: 5 pages,1 figures

Precise weed management is essential for sustaining crop productivity and ecological balance. Traditional herbicide applications face economic and environmental challenges, emphasizing the need for intelligent weed control systems powered by deep learning. These systems require vast amounts of high-quality training data. The reality of scarcity of well-annotated training data, however, is often addressed through generating more data using data augmentation. Nevertheless, conventional augmentation techniques such as random flipping, color changes, and blurring lack sufficient fidelity and diversity. This paper investigates a generative AI-based augmentation technique that uses the Stable Diffusion model to produce diverse synthetic images that improve the quantity and quality of training datasets for weed detection models. Moreover, this paper explores the impact of these synthetic images on the performance of real-time detection systems, thus focusing on compact CNN-based models such as YOLO nano for edge devices. The experimental results show substantial improvements in mean Average Precision (mAP50 and mAP50-95) scores for YOLO models trained with generative AI-augmented datasets, demonstrating the promising potential of synthetic data to enhance model robustness and accuracy.

Multimodal Large Language Models (MLLMs) have made significant strides in visual understanding and generation tasks. However, generating interleaved image-text content remains a challenge, which requires integrated multimodal understanding and generation abilities. While the progress in unified models offers new solutions, existing benchmarks are insufficient for evaluating these methods due to data size and diversity limitations. To bridge this gap, we introduce GATE OpenING (OpenING), a comprehensive benchmark comprising 5,400 high-quality human-annotated instances across 56 real-world tasks. OpenING covers diverse daily scenarios such as travel guide, design, and brainstorming, offering a robust platform for challenging interleaved generation methods. In addition, we present IntJudge, a judge model for evaluating open-ended multimodal generation methods. Trained with a novel data pipeline, our IntJudge achieves an agreement rate of 82. 42% with human judgments, outperforming GPT-based evaluators by 11.34%. Extensive experiments on OpenING reveal that current interleaved generation methods still have substantial room for improvement. Key findings on interleaved image-text generation are further presented to guide the development of next-generation models. The OpenING is open-sourced at https://opening.github.io., Comment: 53 pages, 19 figures

This paper presents a preliminary study of an efficient object tracking approach, comparing the performance of two different 3D point cloud sensory sources: LiDAR and stereo cameras, which have significant price differences. In this preliminary work, we focus on single object tracking. We first developed a fast heuristic object detector that utilizes prior information about the environment and target. The resulting target points are subsequently fed into an extended object tracking framework, where the target shape is parameterized using a star-convex hypersurface model. Experimental results show that our object tracking method using a stereo camera achieves performance similar to that of a LiDAR sensor, with a cost difference of more than tenfold.

Fast progress in 3D Gaussian Splatting (3DGS) has made 3D Gaussians popular for 3D modeling and image rendering, but this creates big challenges in data storage and transmission. To obtain a highly compact 3DGS representation, we propose a hybrid entropy model for Gaussian Splatting (HEMGS) data compression, which comprises two primary components, a hyperprior network and an autoregressive network. To effectively reduce structural redundancy across attributes, we apply a progressive coding algorithm to generate hyperprior features, in which we use previously compressed attributes and location as prior information. In particular, to better extract the location features from these compressed attributes, we adopt a domain-aware and instance-aware architecture to respectively capture domain-aware structural relations without additional storage costs and reveal scene-specific features through MLPs. Additionally, to reduce redundancy within each attribute, we leverage relationships between neighboring compressed elements within the attributes through an autoregressive network. Given its unique structure, we propose an adaptive context coding algorithm with flexible receptive fields to effectively capture adjacent compressed elements. Overall, we integrate our HEMGS into an end-to-end optimized 3DGS compression framework and the extensive experimental results on four benchmarks indicate that our method achieves about 40\% average reduction in size while maintaining the rendering quality over our baseline method and achieving state-of-the-art compression results.

Accurately mapping large-scale cropland is crucial for agricultural production management and planning. Currently, the combination of remote sensing data and deep learning techniques has shown outstanding performance in cropland mapping. However, those approaches require massive precise labels, which are labor-intensive. To reduce the label cost, this study presented a weakly supervised framework considering multi-temporal information for large-scale cropland mapping. Specifically, we extract high-quality labels according to their consistency among global land cover (GLC) products to construct the supervised learning signal. On the one hand, to alleviate the overfitting problem caused by the model's over-trust of remaining errors in high-quality labels, we encode the similarity/aggregation of cropland in the visual/spatial domain to construct the unsupervised learning signal, and take it as the regularization term to constrain the supervised part. On the other hand, to sufficiently leverage the plentiful information in the samples without high-quality labels, we also incorporate the unsupervised learning signal in these samples, enriching the diversity of the feature space. After that, to capture the phenological features of croplands, we introduce dense satellite image time series (SITS) to extend the proposed framework in the temporal dimension. We also visualized the high dimensional phenological features to uncover how multi-temporal information benefits cropland extraction, and assessed the method's robustness under conditions of data scarcity. The proposed framework has been experimentally validated for strong adaptability across three study areas (Hunan Province, Southeast France, and Kansas) in large-scale cropland mapping, and the internal mechanism and temporal generalizability are also investigated.

Recent advancements in all-in-one image restoration models have revolutionized the ability to address diverse degradations through a unified framework. However, parameters tied to specific tasks often remain inactive for other tasks, making mixture-of-experts (MoE) architectures a natural extension. Despite this, MoEs often show inconsistent behavior, with some experts unexpectedly generalizing across tasks while others struggle within their intended scope. This hinders leveraging MoEs' computational benefits by bypassing irrelevant experts during inference. We attribute this undesired behavior to the uniform and rigid architecture of traditional MoEs. To address this, we introduce ``complexity experts" -- flexible expert blocks with varying computational complexity and receptive fields. A key challenge is assigning tasks to each expert, as degradation complexity is unknown in advance. Thus, we execute tasks with a simple bias toward lower complexity. To our surprise, this preference effectively drives task-specific allocation, assigning tasks to experts with the appropriate complexity. Extensive experiments validate our approach, demonstrating the ability to bypass irrelevant experts during inference while maintaining superior performance. The proposed MoCE-IR model outperforms state-of-the-art methods, affirming its efficiency and practical applicability. The source will be publicly made available at \href{https://eduardzamfir.github.io/moceir/}{\texttt{eduardzamfir.github.io/MoCE-IR/}}