CVPR 2024 Papers

Generative Vision-Language Models (VLMs) are prone to generate plausible-sounding textual answers which however are not always grounded in the input image. We investigate this phenomenon usually referred to as "hallucination" and show that it stems from an excessive reliance on the language prior. In particular we show that as more tokens are generated the reliance on the visual prompt decreases and this behavior strongly correlates with the emergence of hallucinations. To reduce hallucinations we introduce Multi-Modal Mutual-Information Decoding (M3ID) a new sampling method for prompt amplification. M3ID amplifies the influence of the reference image over the language prior hence favoring the generation of tokens with higher mutual information with the visual prompt. M3ID can be applied to any pre-trained autoregressive VLM at inference time without necessitating further training and with minimal computational overhead. If training is an option we show that M3ID can be paired with Direct Preference Optimization (DPO) to improve the model's reliance on the prompt image without requiring any labels. Our empirical findings show that our algorithms maintain the fluency and linguistic capabilities of pre-trained VLMs while reducing hallucinations by mitigating visually ungrounded answers. Specifically for the LLaVA 13B model M3ID and M3ID+DPO reduce the percentage of hallucinated objects in captioning tasks by 25% and 28% respectively and improve the accuracy on VQA benchmarks such as POPE by 21% and 24%.

Recently integrating video foundation models and large language models to build a video understanding system can overcome the limitations of specific pre-defined vision tasks. Yet existing systems can only handle videos with very few frames. For long videos the computation complexity memory cost and long-term temporal connection impose additional challenges. Taking advantage of the Atkinson-Shiffrin memory model with tokens in Transformers being employed as the carriers of memory in combination with our specially designed memory mechanism we propose the MovieChat to overcome these challenges. MovieChat achieves state-of-the-art performance in long video understanding along with the released MovieChat-1K benchmark with 1K long video and 14K manual annotations for validation of the effectiveness of our method. The code models and data can be found in https://rese1f.github.io/MovieChat.

Neural rendering has demonstrated remarkable success in dynamic scene reconstruction. Thanks to the expressiveness of neural representations prior works can accurately capture the motion and achieve high-fidelity reconstruction of the target object. Despite this real-world video scenarios often feature large unobserved regions where neural representations struggle to achieve realistic completion. To tackle this challenge we introduce MorpheuS a framework for dynamic 360deg surface reconstruction from a casually captured RGB-D video. Our approach models the target scene as a canonical field that encodes its geometry and appearance in conjunction with a deformation field that warps points from the current frame to the canonical space. We leverage a view-dependent diffusion prior and distill knowledge from it to achieve realistic completion of unobserved regions. Experimental results on various real-world and synthetic datasets show that our method can achieve high-fidelity 360deg surface reconstruction of a deformable object from a monocular RGB-D video.

Recent 3D face reconstruction methods have made remarkable advancements yet there remain huge challenges in monocular high-quality facial reflectance reconstruction. Existing methods rely on a large amount of light-stage captured data to learn facial reflectance models. However the lack of subject diversity poses challenges in achieving good generalization and widespread applicability. In this paper we learn the reflectance prior in image space rather than UV space and present a framework named ID2Reflectance. Our framework can directly estimate the reflectance maps of a single image while using limited reflectance data for training. Our key insight is that reflectance data shares facial structures with RGB faces which enables obtaining expressive facial prior from inexpensive RGB data thus reducing the dependency on reflectance data. We first learn a high-quality prior for facial reflectance. Specifically we pretrain multi-domain facial feature codebooks and design a codebook fusion method to align the reflectance and RGB domains. Then we propose an identity-conditioned swapping module that injects facial identity from the target image into the pre-trained auto-encoder to modify the identity of the source reflectance image. Finally we stitch multi-view swapped reflectance images to obtain renderable assets. Extensive experiments demonstrate that our method exhibits excellent generalization capability and achieves state-of-the-art facial reflectance reconstruction results for in-the-wild faces.

Large Multimodal Models (LMMs) have shown promise in vision-language tasks but struggle with high-resolution input and detailed scene understanding. Addressing these challenges we introduce Monkey to enhance LMM capabilities. Firstly Monkey processes input images by dividing them into uniform patches each matching the size (e.g. 448x448) used in the original training of the well-trained vision encoder. Equipped with individual adapter for each patch Monkey can handle higher resolutions up to 1344x896 pixels enabling the detailed capture of complex visual information. Secondly it employs a multi-level description generation method enriching the context for scene-object associations. This two-part strategy ensures more effective learning from generated data: the higher resolution allows for a more detailed capture of visuals which in turn enhances the effectiveness of comprehensive descriptions. Extensive ablative results validate the effectiveness of our designs. Additionally experiments on 18 datasets further demonstrate that Monkey surpasses existing LMMs in many tasks like Image Captioning and various Visual Question Answering formats. Specially in qualitative tests focused on dense text question answering Monkey has exhibited encouraging results compared with GPT4V. Code is available at https://github.com/Yuliang-Liu/Monkey.

Understanding social interactions involving both verbal and non-verbal cues is essential for effectively interpreting social situations. However most prior works on multimodal social cues focus predominantly on single-person behaviors or rely on holistic visual representations that are not aligned to utterances in multi-party environments. Consequently they are limited in modeling the intricate dynamics of multi-party interactions. In this paper we introduce three new challenging tasks to model the fine-grained dynamics between multiple people: speaking target identification pronoun coreference resolution and mentioned player prediction. We contribute extensive data annotations to curate these new challenges in social deduction game settings. Furthermore we propose a novel multimodal baseline that leverages densely aligned language-visual representations by synchronizing visual features with their corresponding utterances. This facilitates concurrently capturing verbal and non-verbal cues pertinent to social reasoning. Experiments demonstrate the effectiveness of the proposed approach with densely aligned multimodal representations in modeling fine-grained social interactions. Project website: https://sangmin-git.github.io/projects/MMSI.

Integrating whole-slide images (WSIs) and bulk transcriptomics for predicting patient survival can improve our understanding of patient prognosis. However this multimodal task is particularly challenging due to the different nature of these data: WSIs represent a very high-dimensional spatial description of a tumor while bulk transcriptomics represent a global description of gene expression levels within that tumor. In this context our work aims to address two key challenges: (1) how can we tokenize transcriptomics in a semantically meaningful and interpretable way? and (2) how can we capture dense multimodal interactions between these two modalities? Here we propose to learn biological pathway tokens from transcriptomics that can encode specific cellular functions. Together with histology patch tokens that encode the slide morphology we argue that they form appropriate reasoning units for interpretability. We fuse both modalities using a memory-efficient multimodal Transformer that can model interactions between pathway and histology patch tokens. Our model SURVPATH achieves state-of-the-art performance when evaluated against unimodal and multimodal baselines on five datasets from The Cancer Genome Atlas. Our interpretability framework identifies key multimodal prognostic factors and as such can provide valuable insights into the interaction between genotype and phenotype. Code available at https://github.com/mahmoodlab/SurvPath.

From content moderation to wildlife conservation the number of applications that require models to recognize nuanced or subjective visual concepts is growing. Traditionally developing classifiers for such concepts requires substantial manual effort measured in hours days or even months to identify and annotate data needed for training. Even with recently proposed Agile Modeling techniques which enable rapid bootstrapping of image classifiers users are still required to spend 30 minutes or more of monotonous repetitive data labeling just to train a single classifier. Drawing on Fiske's Cognitive Miser theory we propose a new framework that alleviates manual effort by replacing human labeling with natural language interactions reducing the total effort required to define a concept by an order of magnitude: from labeling 2000 images to only 100 plus some natural language interactions. Our framework leverages recent advances in foundation models both large language models and vision-language models to carve out the concept space through conversation and by automatically labeling training data points. Most importantly our framework eliminates the need for crowd-sourced annotations. Moreover our framework ultimately produces lightweight classification models that are deployable in cost-sensitive scenarios. Across 15 subjective concepts and across 2 public image classification datasets our trained models outperform traditional Agile Modeling as well as state-of-the-art zero-shot classification models like ALIGN CLIP CuPL and large visual question answering models like PaLI-X.

Video-based Unsupervised Domain Adaptation (VUDA) method improves the generalization of the video model enabling it to be applied to action recognition tasks in different environments. However these methods require continuous access to source data during the adaptation process which are impractical in real scenarios where the source videos are not available with concerns in transmission efficiency or privacy issues. To address this problem in this paper we propose to solve the Multimodal Video Test-Time Adaptation task (MVTTA). Existing image-based TTA methods cannot be directly applied to this task because video have domain shift in multimodal and temporal which brings difficulties to adaptation. To address the above challenges we propose a Modality-Collaborative Test-Time Adaptation (MC-TTA) Network. We maintain teacher and student memory banks respectively for generating pseudo-prototypes and target-prototypes. In the teacher model we propose Self-assembled Source-friendly Feature Reconstruction (SSFR) module to encourage the teacher memory bank to store features that are more likely to be consistent with the source distribution. Through multimodal prototype alignment and cross-modal relative consistency our method can effectively alleviate domain shift in videos. We evaluate the proposed model on four public video datasets. The results show that our model outperforms existing state-of-the-art methods.

Establishing dense anatomical correspondence across distinct imaging modalities is a foundational yet challenging procedure for numerous medical image analysis studies and image-guided radiotherapy. Existing multi-modality image registration algorithms rely on statistical-based similarity measures or local structural image representations. However the former is sensitive to locally varying noise while the latter is not discriminative enough to cope with complex anatomical structures in multimodal scans causing ambiguity in determining the anatomical correspondence across scans with different modalities. In this paper we propose a modality-agnostic structural representation learning method which leverages Deep Neighbourhood Self-similarity (DNS) and anatomy-aware contrastive learning to learn discriminative and contrast-invariance deep structural image representations (DSIR) without the need for anatomical delineations or pre-aligned training images. We evaluate our method on multiphase CT abdomen MR-CT and brain MR T1w-T2w registration. Comprehensive results demonstrate that our method is superior to the conventional local structural representation and statistical-based similarity measures in terms of discriminability and accuracy.

Humans possess the capability to comprehend diverse modalities and seamlessly transfer information between them. In this work we introduce ModaVerse a Multi-modal Large Language Model (MLLM) capable of comprehending and transforming content across various modalities including images videos and audio. Predominant MLLM frameworks have largely relied on aligning latent spaces of textual and non-textual features. This alignment process which synchronizes a language model trained on textual data with encoders and decoders trained on multi-modal data often necessitates extensive training of several projection layers in multiple stages. Inspired by LLM-as-agent methodologies we propose a novel Input/Output (I/O) alignment mechanism that operates directly at the level of natural language. It aligns the LLM's output with the input of generative models avoiding the complexities associated with latent feature alignments and simplifying the multiple training stages of existing MLLMs into a single efficient process. By conducting experiments on several benchmarks we demonstrate that our approach attains comparable performance with the state of the art while achieving considerable efficiencies in data usage.

Vision-language pre-trained models have achieved impressive performance on various downstream tasks. However their large model sizes hinder their utilization on platforms with limited computational resources. We find that directly using smaller pre-trained models and applying magnitude-based pruning on CLIP models leads to inflexibility and inferior performance. Recent efforts for VLP compression either adopt uni-modal compression metrics resulting in limited performance or involve costly mask-search processes with learnable masks. In this paper we first propose the Module-wise Pruning Error (MoPE) metric accurately assessing CLIP module importance by performance decline on cross-modal tasks. Using the MoPE metric we introduce a unified pruning framework applicable to both pre-training and task-specific fine-tuning compression stages. For pre-training MoPE-CLIP effectively leverages knowledge from the teacher model significantly reducing pre-training costs while maintaining strong zero-shot capabilities. For fine-tuning consecutive pruning from width to depth yields highly competitive task-specific models. Extensive experiments in two stages demonstrate the effectiveness of the MoPE metric and MoPE-CLIP outperforms previous state-of-the-art VLP compression methods.

Learning-based stereo matching techniques have made significant progress. However existing methods inevitably lose geometrical structure information during the feature channel generation process resulting in edge detail mismatches. In this paper the Motif Channel Attention Stereo Matching Network (MoCha-Stereo) is designed to address this problem. We provide the Motif Channel Correlation Volume (MCCV) to determine more accurate edge matching costs. MCCV is achieved by projecting motif channels which capture common geometric structures in feature channels onto feature maps and cost volumes. In addition edge variations in the reconstruction error map also affect details matching we propose the Reconstruction Error Motif Penalty (REMP) module to further refine the full-resolution disparity estimation. REMP integrates the frequency information of typical channel features from the reconstruction error. MoCha-Stereo ranks 1st on the KITTI-2015 and KITTI-2012 Reflective leaderboards. Our structure also shows excellent performance in Multi-View Stereo. Code is avaliable at https://github.com/ZYangChen/MoCha-Stereo.

Large Vision-Language Models (LVLMs) have advanced considerably intertwining visual recognition and language understanding to generate content that is not only coherent but also contextually attuned. Despite their success LVLMs still suffer from the issue of object hallucinations where models generate plausible yet incorrect outputs that include objects that do not exist in the images. To mitigate this issue we introduce Visual Contrastive Decoding (VCD) a simple and training-free method that contrasts output distributions derived from original and distorted visual inputs. The proposed VCD effectively reduces the over-reliance on statistical bias and unimodal priors two essential causes of object hallucinations. This adjustment ensures the generated content is closely grounded to visual inputs resulting in contextually accurate outputs. Our experiments show that VCD without either additional training or the usage of external tools significantly mitigates the object hallucination issue across different LVLM families. Beyond mitigating object hallucinations VCD also excels in general LVLM benchmarks highlighting its wide-ranging applicability.

Noisy correspondence that refers to mismatches in cross-modal data pairs is prevalent on human-annotated or web-crawled datasets. Prior approaches to leverage such data mainly consider the application of uni-modal noisy label learning without amending the impact on both cross-modal and intra-modal geometrical structures in multimodal learning. Actually we find that both structures are effective to discriminate noisy correspondence through structural differences when being well-established. Inspired by this observation we introduce a Geometrical Structure Consistency (GSC) method to infer the true correspondence. Specifically GSC ensures the preservation of geometrical structures within and between modalities allowing for the accurate discrimination of noisy samples based on structural differences. Utilizing these inferred true correspondence labels GSC refines the learning of geometrical structures by filtering out the noisy samples. Experiments across four cross-modal datasets confirm that GSC effectively identifies noisy samples and significantly outperforms the current leading methods. Source code is available at https://github.com/MediaBrain-SJTU/GSC.

This paper aims to address a common challenge in deep learning-based image transformation methods such as image enhancement and super-resolution which heavily rely on precisely aligned paired datasets with pixel-level alignments. However creating precisely aligned paired images presents significant challenges and hinders the advancement of methods trained on such data. To overcome this challenge this paper introduces a novel and simple Frequency Distribution Loss (FDL) for computing distribution distance within the frequency domain. Specifically we transform image features into the frequency domain using Discrete Fourier Transformation (DFT). Subsequently frequency components (amplitude and phase) are processed separately to form the FDL loss function. Our method is empirically proven effective as a training constraint due to the thoughtful utilization of global information in the frequency domain. Extensive experimental evaluations focusing on image enhancement and super-resolution tasks demonstrate that FDL outperforms existing misalignment-robust loss functions. Furthermore we explore the potential of our FDL for image style transfer that relies solely on completely misaligned data. Our code is available at: https://github.com/eezkni/FDL

One of the main challenges of multimodal learning is the need to combine heterogeneous modalities (e.g. video audio text). For example video and audio are obtained at much higher rates than text and are roughly aligned in time. They are often not synchronized with text which comes as a global context e.g. a title or a description. Furthermore video and audio inputs are of much larger volumes and grow as the video length increases which naturally requires more compute dedicated to these modalities and makes modeling of long-range dependencies harder. We here decouple the multimodal modeling dividing it into separate autoregressive models processing the inputs according to the characteristics of the modalities. We propose a multimodal model consisting of an autoregressive component for the time-synchronized modalities (audio and video) and an autoregressive component for the context modalities which are not necessarily aligned in time but are still sequential. To address the long-sequences of the video-audio inputs we further partition the video and audio sequences in consecutive snippets and autoregressively process their representations. To that end we propose a Combiner mechanism which models the audio-video information jointly producing compact but expressive representations. This allows us to scale to 512 input video frames without increase in model parameters. Our approach achieves the state-of-the-art on multiple well established multimodal benchmarks. It effectively addresses the high computational demand of media inputs by learning compact representations controlling the sequence length of the audio-video feature representations and modeling their dependencies in time.

We introduce a new family of minimal problems for reconstruction from multiple views. Our primary focus is a novel approach to autocalibration a long-standing problem in computer vision. Traditional approaches to this problem such as those based on Kruppa's equations or the modulus constraint rely explicitly on the knowledge of multiple fundamental matrices or a projective reconstruction. In contrast we consider a novel formulation involving constraints on image points the unknown depths of 3D points and a partially specified calibration matrix K. For 2 and 3 views we present a comprehensive taxonomy of minimal autocalibration problems obtained by relaxing some of these constraints. These problems are organized into classes according to the number of views and any assumed prior knowledge of K. Within each class we determine problems with the fewest---or a relatively small number of---solutions. From this zoo of problems we devise three practical solvers. Experiments with synthetic and real data and interfacing our solvers with COLMAP demonstrate that we achieve superior accuracy compared to state-of-the-art calibration methods. The code is available at https://github.com/andreadalcin/MinimalPerspectiveAutocalibration.

Category-agnostic pose estimation (CAPE) aims to predict keypoints for arbitrary classes given a few support images annotated with keypoints. Existing methods only rely on the features extracted at support keypoints to predict or refine the keypoints on query image but a few support feature vectors are local and inadequate for CAPE. Considering that human can quickly perceive potential keypoints of arbitrary objects we propose a novel framework for CAPE based on such potential keypoints (named as meta-points). Specifically we maintain learnable embeddings to capture inherent information of various keypoints which interact with image feature maps to produce meta-points without any support. The produced meta-points could serve as meaningful potential keypoints for CAPE. Due to the inevitable gap between inherency and annotation we finally utilize the identities and details offered by support keypoints to assign and refine meta-points to desired keypoints in query image. In addition we propose a progressive deformable point decoder and a slacked regression loss for better prediction and supervision. Our novel framework not only reveals the inherency of keypoints but also outperforms existing methods of CAPE. Comprehensive experiments and in-depth studies on large-scale MP-100 dataset demonstrate the effectiveness of our framework.