PyTorchKR Papers

최신 논문

총 139,333편
태그·기간 등 상세 필터

카테고리 탐색

arXiv:2505.13215공식 코드 · Papers with Code

Hybrid 3D-4D Gaussian Splatting for Fast Dynamic Scene Representation

Recent advancements in dynamic 3D scene reconstruction have shown promising results, enabling high-fidelity 3D novel view synthesis with improved temporal consistency. Among these, 4D Gaussian Splatting (4DGS) has emerged as an appealing approach due to its ability to model high-fidelity spatial and temporal variations. However, existing methods suffer from substantial computational and memory overhead due to the redundant allocation of 4D Gaussians to static regions, which can also degrade image quality. In this work, we introduce hybrid 3D-4D Gaussian Splatting (3D-4DGS), a novel framework that adaptively represents static regions with 3D Gaussians while reserving 4D Gaussians for dynamic elements. Our method begins with a fully 4D Gaussian representation and iteratively converts temporally invariant Gaussians into 3D, significantly reducing the number of parameters and improving computational efficiency. Meanwhile, dynamic Gaussians retain their full 4D representation, capturing complex motions with high fidelity. Our approach achieves significantly faster training times compared to baseline 4D Gaussian Splatting methods while maintaining or improving the visual quality.

arXiv:2505.13152공식 코드 · Papers with Code

Higher fidelity perceptual image and video compression with a latent conditioned residual denoising diffusion model

Denoising diffusion models achieved impressive results on several image generation tasks often outperforming GAN based models. Recently, the generative capabilities of diffusion models have been employed for perceptual image compression, such as in CDC. A major drawback of these diffusion-based methods is that, while producing impressive perceptual quality images they are dropping in fidelity/increasing the distortion to the original uncompressed images when compared with other traditional or learned image compression schemes aiming for fidelity. In this paper, we propose a hybrid compression scheme optimized for perceptual quality, extending the approach of the CDC model with a decoder network in order to reduce the impact on distortion metrics such as PSNR. After using the decoder network to generate an initial image, optimized for distortion, the latent conditioned diffusion model refines the reconstruction for perceptual quality by predicting the residual. On standard benchmarks, we achieve up to +2dB PSNR fidelity improvements while maintaining comparable LPIPS and FID perceptual scores when compared with CDC. Additionally, the approach is easily extensible to video compression, where we achieve similar results.

arXiv:2505.13137공식 코드 · Papers with Code

Learning to Adapt to Position Bias in Vision Transformer Classifiers

How discriminative position information is for image classification depends on the data. On the one hand, the camera position is arbitrary and objects can appear anywhere in the image, arguing for translation invariance. At the same time, position information is key for exploiting capture/center bias, and scene layout, e.g.: the sky is up. We show that position bias, the level to which a dataset is more easily solved when positional information on input features is used, plays a crucial role in the performance of Vision Transformers image classifiers. To investigate, we propose Position-SHAP, a direct measure of position bias by extending SHAP to work with position embeddings. We show various levels of position bias in different datasets, and find that the optimal choice of position embedding depends on the position bias apparent in the dataset. We therefore propose Auto-PE, a single-parameter position embedding extension, which allows the position embedding to modulate its norm, enabling the unlearning of position information. Auto-PE combines with existing PEs to match or improve accuracy on classification datasets.

arXiv:2505.13088공식 코드 · Papers with Code

Cross-modal feature fusion for robust point cloud registration with ambiguous geometry

Point cloud registration has seen significant advancements with the application of deep learning techniques. However, existing approaches often overlook the potential of integrating radiometric information from RGB images. This limitation reduces their effectiveness in aligning point clouds pairs, especially in regions where geometric data alone is insufficient. When used effectively, radiometric information can enhance the registration process by providing context that is missing from purely geometric data. In this paper, we propose CoFF, a novel Cross-modal Feature Fusion method that utilizes both point cloud geometry and RGB images for pairwise point cloud registration. Assuming that the co-registration between point clouds and RGB images is available, CoFF explicitly addresses the challenges where geometric information alone is unclear, such as in regions with symmetric similarity or planar structures, through a two-stage fusion of 3D point cloud features and 2D image features. It incorporates a cross-modal feature fusion module that assigns pixel-wise image features to 3D input point clouds to enhance learned 3D point features, and integrates patch-wise image features with superpoint features to improve the quality of coarse matching. This is followed by a coarse-to-fine matching module that accurately establishes correspondences using the fused features. We extensively evaluate CoFF on four common datasets: 3DMatch, 3DLoMatch, IndoorLRS, and the recently released ScanNet++ datasets. In addition, we assess CoFF on specific subset datasets containing geometrically ambiguous cases. Our experimental results demonstrate that CoFF achieves state-of-the-art registration performance across all benchmarks, including remarkable registration recalls of 95.9% and 81.6% on the widely-used 3DMatch and 3DLoMatch datasets, respectively...(Truncated to fit arXiv abstract length)

arXiv:2505.13061공식 코드 · Papers with Code

3D Visual Illusion Depth Estimation

3D visual illusion is a perceptual phenomenon where a two-dimensional plane is manipulated to simulate three-dimensional spatial relationships, making a flat artwork or object look three-dimensional in the human visual system. In this paper, we reveal that the machine visual system is also seriously fooled by 3D visual illusions, including monocular and binocular depth estimation. In order to explore and analyze the impact of 3D visual illusion on depth estimation, we collect a large dataset containing almost 3k scenes and 200k images to train and evaluate SOTA monocular and binocular depth estimation methods. We also propose a robust depth estimation framework that uses common sense from a vision-language model to adaptively select reliable depth from binocular disparity and monocular depth. Experiments show that SOTA monocular, binocular, and multi-view depth estimation approaches are all fooled by various 3D visual illusions, while our method achieves SOTA performance.

arXiv:2505.12999공식 코드 · Papers with Code

A generalisable head MRI defacing pipeline: Evaluation on 2,566 meningioma scans

Reliable MRI defacing techniques to safeguard patient privacy while preserving brain anatomy are critical for research collaboration. Existing methods often struggle with incomplete defacing or degradation of brain tissue regions. We present a robust, generalisable defacing pipeline for high-resolution MRI that integrates atlas-based registration with brain masking. Our method was evaluated on 2,566 heterogeneous clinical scans for meningioma and achieved a 99.92 per cent success rate (2,564/2,566) upon visual inspection. Excellent anatomical preservation is demonstrated with a Dice similarity coefficient of 0.9975 plus or minus 0.0023 between brain masks automatically extracted from the original and defaced volumes. Source code is available at https://github.com/cai4cai/defacing_pipeline.

arXiv:2505.12998공식 코드 · Papers with Code

A Skull-Adaptive Framework for AI-Based 3D Transcranial Focused Ultrasound Simulation

Transcranial focused ultrasound (tFUS) is an emerging modality for non-invasive brain stimulation and therapeutic intervention, offering millimeter-scale spatial precision and the ability to target deep brain structures. However, the heterogeneous and anisotropic nature of the human skull introduces significant distortions to the propagating ultrasound wavefront, which require time-consuming patient-specific planning and corrections using numerical solvers for accurate targeting. To enable data-driven approaches in this domain, we introduce TFUScapes, the first large-scale, high-resolution dataset of tFUS simulations through anatomically realistic human skulls derived from T1-weighted MRI images. We have developed a scalable simulation engine pipeline using the k-Wave pseudo-spectral solver, where each simulation returns a steady-state pressure field generated by a focused ultrasound transducer placed at realistic scalp locations. In addition to the dataset, we present DeepTFUS, a deep learning model that estimates normalized pressure fields directly from input 3D CT volumes and transducer position. The model extends a U-Net backbone with transducer-aware conditioning, incorporating Fourier-encoded position embeddings and MLP layers to create global transducer embeddings. These embeddings are fused with U-Net encoder features via feature-wise modulation, dynamic convolutions, and cross-attention mechanisms. The model is trained using a combination of spatially weighted and gradient-sensitive loss functions, enabling it to approximate high-fidelity wavefields. The TFUScapes dataset is publicly released to accelerate research at the intersection of computational acoustics, neurotechnology, and deep learning. The project page is available at https://github.com/CAMMA-public/TFUScapes.

arXiv:2505.12911공식 코드 · Papers with Code

HiERO: understanding the hierarchy of human behavior enhances reasoning on egocentric videos

Human activities are particularly complex and variable, and this makes challenging for deep learning models to reason about them. However, we note that such variability does have an underlying structure, composed of a hierarchy of patterns of related actions. We argue that such structure can emerge naturally from unscripted videos of human activities, and can be leveraged to better reason about their content. We present HiERO, a weakly-supervised method to enrich video segments features with the corresponding hierarchical activity threads. By aligning video clips with their narrated descriptions, HiERO infers contextual, semantic and temporal reasoning with an hierarchical architecture. We prove the potential of our enriched features with multiple video-text alignment benchmarks (EgoMCQ, EgoNLQ) with minimal additional training, and in zero-shot for procedure learning tasks (EgoProceL and Ego4D Goal-Step). Notably, HiERO achieves state-of-the-art performance in all the benchmarks, and for procedure learning tasks it outperforms fully-supervised methods by a large margin (+12.5% F1 on EgoProceL) in zero shot. Our results prove the relevance of using knowledge of the hierarchy of human activities for multiple reasoning tasks in egocentric vision.

arXiv:2505.12897공식 코드 · Papers with Code

EPIC: Explanation of Pretrained Image Classification Networks via Prototype

Explainable AI (XAI) methods generally fall into two categories. Post-hoc approaches generate explanations for pre-trained models and are compatible with various neural network architectures. These methods often use feature importance visualizations, such as saliency maps, to indicate which input regions influenced the model's prediction. Unfortunately, they typically offer a coarse understanding of the model's decision-making process. In contrast, ante-hoc (inherently explainable) methods rely on specially designed model architectures trained from scratch. A notable subclass of these methods provides explanations through prototypes, representative patches extracted from the training data. However, prototype-based approaches have limitations: they require dedicated architectures, involve specialized training procedures, and perform well only on specific datasets. In this work, we propose EPIC (Explanation of Pretrained Image Classification), a novel approach that bridges the gap between these two paradigms. Like post-hoc methods, EPIC operates on pre-trained models without architectural modifications. Simultaneously, it delivers intuitive, prototype-based explanations inspired by ante-hoc techniques. To the best of our knowledge, EPIC is the first post-hoc method capable of fully replicating the core explanatory power of inherently interpretable models. We evaluate EPIC on benchmark datasets commonly used in prototype-based explanations, such as CUB-200-2011 and Stanford Cars, alongside large-scale datasets like ImageNet, typically employed by post-hoc methods. EPIC uses prototypes to explain model decisions, providing a flexible and easy-to-understand tool for creating clear, high-quality explanations.

arXiv:2505.12861공식 코드 · Papers with Code

Robust Multimodal Segmentation with Representation Regularization and Hybrid Prototype Distillation

Multi-modal semantic segmentation (MMSS) faces significant challenges in real-world scenarios due to dynamic environments, sensor failures, and noise interference, creating a gap between theoretical models and practical performance. To address this, we propose a two-stage framework called RobustSeg, which enhances multi-modal robustness through two key components: the Hybrid Prototype Distillation Module (HPDM) and the Representation Regularization Module (RRM). In the first stage, RobustSeg pre-trains a multi-modal teacher model using complete modalities. In the second stage, a student model is trained with random modality dropout while learning from the teacher via HPDM and RRM. HPDM transforms features into compact prototypes, enabling cross-modal hybrid knowledge distillation and mitigating bias from missing modalities. RRM reduces representation discrepancies between the teacher and student by optimizing functional entropy through the log-Sobolev inequality. Extensive experiments on three public benchmarks demonstrate that RobustSeg outperforms previous state-of-the-art methods, achieving improvements of +2.76%, +4.56%, and +0.98%, respectively. Code is available at: https://github.com/RobustSeg/RobustSeg.

arXiv:2505.12849공식 코드 · Papers with Code

Accelerate TarFlow Sampling with GS-Jacobi Iteration

Image generation models have achieved widespread applications. As an instance, the TarFlow model combines the transformer architecture with Normalizing Flow models, achieving state-of-the-art results on multiple benchmarks. However, due to the causal form of attention requiring sequential computation, TarFlow's sampling process is extremely slow. In this paper, we demonstrate that through a series of optimization strategies, TarFlow sampling can be greatly accelerated by using the Gauss-Seidel-Jacobi (abbreviated as GS-Jacobi) iteration method. Specifically, we find that blocks in the TarFlow model have varying importance: a small number of blocks play a major role in image generation tasks, while other blocks contribute relatively little; some blocks are sensitive to initial values and prone to numerical overflow, while others are relatively robust. Based on these two characteristics, we propose the Convergence Ranking Metric (CRM) and the Initial Guessing Metric (IGM): CRM is used to identify whether a TarFlow block is "simple" (converges in few iterations) or "tough" (requires more iterations); IGM is used to evaluate whether the initial value of the iteration is good. Experiments on four TarFlow models demonstrate that GS-Jacobi sampling can significantly enhance sampling efficiency while maintaining the quality of generated images (measured by FID), achieving speed-ups of 4.53x in Img128cond, 5.32x in AFHQ, 2.96x in Img64uncond, and 2.51x in Img64cond without degrading FID scores or sample quality. Code and checkpoints are accessible on https://github.com/encoreus/GS-Jacobi_for_TarFlow

arXiv:2505.12835공식 코드 · Papers with Code

FlightGPT: Towards Generalizable and Interpretable UAV Vision-and-Language Navigation with Vision-Language Models

Unmanned Aerial Vehicle (UAV) Vision-and-Language Navigation (VLN) is vital for applications such as disaster response, logistics delivery, and urban inspection. However, existing methods often struggle with insufficient multimodal fusion, weak generalization, and poor interpretability. To address these challenges, we propose FlightGPT, a novel UAV VLN framework built upon Vision-Language Models (VLMs) with powerful multimodal perception capabilities. We design a two-stage training pipeline: first, Supervised Fine-Tuning (SFT) using high-quality demonstrations to improve initialization and structured reasoning; then, Group Relative Policy Optimization (GRPO) algorithm, guided by a composite reward that considers goal accuracy, reasoning quality, and format compliance, to enhance generalization and adaptability. Furthermore, FlightGPT introduces a Chain-of-Thought (CoT)-based reasoning mechanism to improve decision interpretability. Extensive experiments on the city-scale dataset CityNav demonstrate that FlightGPT achieves state-of-the-art performance across all scenarios, with a 9.22\% higher success rate than the strongest baseline in unseen environments. Our implementation is publicly available.

arXiv:2505.12820공식 코드 · Papers with Code

Rethinking Features-Fused-Pyramid-Neck for Object Detection

Multi-head detectors typically employ a features-fused-pyramid-neck for multi-scale detection and are widely adopted in the industry. However, this approach faces feature misalignment when representations from different hierarchical levels of the feature pyramid are forcibly fused point-to-point. To address this issue, we designed an independent hierarchy pyramid (IHP) architecture to evaluate the effectiveness of the features-unfused-pyramid-neck for multi-head detectors. Subsequently, we introduced soft nearest neighbor interpolation (SNI) with a weight downscaling factor to mitigate the impact of feature fusion at different hierarchies while preserving key textures. Furthermore, we present a features adaptive selection method for down sampling in extended spatial windows (ESD) to retain spatial features and enhance lightweight convolutional techniques (GSConvE). These advancements culminate in our secondary features alignment solution (SA) for real-time detection, achieving state-of-the-art results on Pascal VOC and MS COCO. Code will be released at https://github.com/AlanLi1997/rethinking-fpn. This paper has been accepted by ECCV2024 and published on Springer Nature.

arXiv:2505.12766공식 코드 · Papers with Code

Reasoning-OCR: Can Large Multimodal Models Solve Complex Logical Reasoning Problems from OCR Cues?

Large Multimodal Models (LMMs) have become increasingly versatile, accompanied by impressive Optical Character Recognition (OCR) related capabilities. Existing OCR-related benchmarks emphasize evaluating LMMs' abilities of relatively simple visual question answering, visual-text parsing, etc. However, the extent to which LMMs can deal with complex logical reasoning problems based on OCR cues is relatively unexplored. To this end, we introduce the Reasoning-OCR benchmark, which challenges LMMs to solve complex reasoning problems based on the cues that can be extracted from rich visual-text. Reasoning-OCR covers six visual scenarios and encompasses 150 meticulously designed questions categorized into six reasoning challenges. Additionally, Reasoning-OCR minimizes the impact of field-specialized knowledge. Our evaluation offers some insights for proprietary and open-source LMMs in different reasoning challenges, underscoring the urgent to improve the reasoning performance. We hope Reasoning-OCR can inspire and facilitate future research on enhancing complex reasoning ability based on OCR cues. Reasoning-OCR is publicly available at https://github.com/Hxyz-123/ReasoningOCR.

arXiv:2505.12742공식 코드 · Papers with Code

MVAR: Visual Autoregressive Modeling with Scale and Spatial Markovian Conditioning

Essential to visual generation is efficient modeling of visual data priors. Conventional next-token prediction methods define the process as learning the conditional probability distribution of successive tokens. Recently, next-scale prediction methods redefine the process to learn the distribution over multi-scale representations, significantly reducing generation latency. However, these methods condition each scale on all previous scales and require each token to consider all preceding tokens, exhibiting scale and spatial redundancy. To better model the distribution by mitigating redundancy, we propose Markovian Visual AutoRegressive modeling (MVAR), a novel autoregressive framework that introduces scale and spatial Markov assumptions to reduce the complexity of conditional probability modeling. Specifically, we introduce a scale-Markov trajectory that only takes as input the features of adjacent preceding scale for next-scale prediction, enabling the adoption of a parallel training strategy that significantly reduces GPU memory consumption. Furthermore, we propose spatial-Markov attention, which restricts the attention of each token to a localized neighborhood of size k at corresponding positions on adjacent scales, rather than attending to every token across these scales, for the pursuit of reduced modeling complexity. Building on these improvements, we reduce the computational complexity of attention calculation from O(N^2) to O(Nk), enabling training with just eight NVIDIA RTX 4090 GPUs and eliminating the need for KV cache during inference. Extensive experiments on ImageNet demonstrate that MVAR achieves comparable or superior performance with both small model trained from scratch and large fine-tuned models, while reducing the average GPU memory footprint by 3.0x.

arXiv:2505.12728공식 코드 · Papers with Code

FLASH: Latent-Aware Semi-Autoregressive Speculative Decoding for Multimodal Tasks

Large language and multimodal models (LLMs and LMMs) exhibit strong inference capabilities but are often limited by slow decoding speeds. This challenge is especially acute in LMMs, where visual inputs typically comprise more tokens with lower information density than text -- an issue exacerbated by recent trends toward finer-grained visual tokenizations to boost performance. Speculative decoding has been effective in accelerating LLM inference by using a smaller draft model to generate candidate tokens, which are then selectively verified by the target model, improving speed without sacrificing output quality. While this strategy has been extended to LMMs, existing methods largely overlook the unique properties of visual inputs and depend solely on text-based draft models. In this work, we propose \textbf{FLASH} (Fast Latent-Aware Semi-Autoregressive Heuristics), a speculative decoding framework designed specifically for LMMs, which leverages two key properties of multimodal data to design the draft model. First, to address redundancy in visual tokens, we propose a lightweight latent-aware token compression mechanism. Second, recognizing that visual objects often co-occur within a scene, we employ a semi-autoregressive decoding strategy to generate multiple tokens per forward pass. These innovations accelerate draft decoding while maintaining high acceptance rates, resulting in faster overall inference. Experiments show that FLASH significantly outperforms prior speculative decoding approaches in both unimodal and multimodal settings, achieving up to \textbf{2.68$\times$} speed-up on video captioning and \textbf{2.55$\times$} on visual instruction tuning tasks compared to the original LMM. Our code is available \href{https://github.com/ZihuaEvan/FlashSD/}{[here]}.

arXiv:2505.12674공식 코드 · Papers with Code

Few-Step Diffusion via Score identity Distillation

Diffusion distillation has emerged as a promising strategy for accelerating text-to-image (T2I) diffusion models by distilling a pretrained score network into a one- or few-step generator. While existing methods have made notable progress, they often rely on real or teacher-synthesized images to perform well when distilling high-resolution T2I diffusion models such as Stable Diffusion XL (SDXL), and their use of classifier-free guidance (CFG) introduces a persistent trade-off between text-image alignment and generation diversity. We address these challenges by optimizing Score identity Distillation (SiD) -- a data-free, one-step distillation framework -- for few-step generation. Backed by theoretical analysis that justifies matching a uniform mixture of outputs from all generation steps to the data distribution, our few-step distillation algorithm avoids step-specific networks and integrates seamlessly into existing pipelines, achieving state-of-the-art performance on SDXL at 1024x1024 resolution. To mitigate the alignment-diversity trade-off when real text-image pairs are available, we introduce a Diffusion GAN-based adversarial loss applied to the uniform mixture and propose two new guidance strategies: Zero-CFG, which disables CFG in the teacher and removes text conditioning in the fake score network, and Anti-CFG, which applies negative CFG in the fake score network. This flexible setup improves diversity without sacrificing alignment. Comprehensive experiments on SD1.5 and SDXL demonstrate state-of-the-art performance in both one-step and few-step generation settings, along with robustness to the absence of real images. Our efficient PyTorch implementation, along with the resulting one- and few-step distilled generators, will be released publicly as a separate branch at https://github.com/mingyuanzhou/SiD-LSG.

arXiv:2505.12631공식 코드 · Papers with Code

Multi-Resolution Haar Network: Enhancing human motion prediction via Haar transform

The 3D human pose is vital for modern computer vision and computer graphics, and its prediction has drawn attention in recent years. 3D human pose prediction aims at forecasting a human's future motion from the previous sequence. Ignoring that the arbitrariness of human motion sequences has a firm origin in transition in both temporal and spatial axes limits the performance of state-of-the-art methods, leading them to struggle with making precise predictions on complex cases, e.g., arbitrarily posing or greeting. To alleviate this problem, a network called HaarMoDic is proposed in this paper, which utilizes the 2D Haar transform to project joints to higher resolution coordinates where the network can access spatial and temporal information simultaneously. An ablation study proves that the significant contributing module within the HaarModic Network is the Multi-Resolution Haar (MR-Haar) block. Instead of mining in one of two axes or extracting separately, the MR-Haar block projects whole motion sequences to a mixed-up coordinate in higher resolution with 2D Haar Transform, allowing the network to give scope to information from both axes in different resolutions. With the MR-Haar block, the HaarMoDic network can make predictions referring to a broader range of information. Experimental results demonstrate that HaarMoDic surpasses state-of-the-art methods in every testing interval on the Human3.6M dataset in the Mean Per Joint Position Error (MPJPE) metric.

arXiv:2505.12620공식 코드 · Papers with Code

BusterX: MLLM-Powered AI-Generated Video Forgery Detection and Explanation

Advances in AI generative models facilitate super-realistic video synthesis, amplifying misinformation risks via social media and eroding trust in digital content. Several research works have explored new deepfake detection methods on AI-generated images to alleviate these risks. However, with the fast development of video generation models, such as Sora and WanX, there is currently a lack of large-scale, high-quality AI-generated video datasets for forgery detection. In addition, existing detection approaches predominantly treat the task as binary classification, lacking explainability in model decision-making and failing to provide actionable insights or guidance for the public. To address these challenges, we propose \textbf{GenBuster-200K}, a large-scale AI-generated video dataset featuring 200K high-resolution video clips, diverse latest generative techniques, and real-world scenes. We further introduce \textbf{BusterX}, a novel AI-generated video detection and explanation framework leveraging multimodal large language model (MLLM) and reinforcement learning for authenticity determination and explainable rationale. To our knowledge, GenBuster-200K is the {\it \textbf{first}} large-scale, high-quality AI-generated video dataset that incorporates the latest generative techniques for real-world scenarios. BusterX is the {\it \textbf{first}} framework to integrate MLLM with reinforcement learning for explainable AI-generated video detection. Extensive comparisons with state-of-the-art methods and ablation studies validate the effectiveness and generalizability of BusterX. The code, models, and datasets will be released.

arXiv:2505.12051공식 코드 · Papers with Code

Enhanced Multimodal Hate Video Detection via Channel-wise and Modality-wise Fusion

The rapid rise of video content on platforms such as TikTok and YouTube has transformed information dissemination, but it has also facilitated the spread of harmful content, particularly hate videos. Despite significant efforts to combat hate speech, detecting these videos remains challenging due to their often implicit nature. Current detection methods primarily rely on unimodal approaches, which inadequately capture the complementary features across different modalities. While multimodal techniques offer a broader perspective, many fail to effectively integrate temporal dynamics and modality-wise interactions essential for identifying nuanced hate content. In this paper, we present CMFusion, an enhanced multimodal hate video detection model utilizing a novel Channel-wise and Modality-wise Fusion Mechanism. CMFusion first extracts features from text, audio, and video modalities using pre-trained models and then incorporates a temporal cross-attention mechanism to capture dependencies between video and audio streams. The learned features are then processed by channel-wise and modality-wise fusion modules to obtain informative representations of videos. Our extensive experiments on a real-world dataset demonstrate that CMFusion significantly outperforms five widely used baselines in terms of accuracy, precision, recall, and F1 score. Comprehensive ablation studies and parameter analyses further validate our design choices, highlighting the model's effectiveness in detecting hate videos. The source codes will be made publicly available at https://github.com/EvelynZ10/cmfusion.