CVPR 2024 Papers

In this paper we focus on a challenging Online Task-Free Class Incremental Learning (OTFCIL) problem. Different from the existing methods that continuously learn the feature space from data streams we propose a novel compute-and-align paradigm for the OTFCIL. It first computes an optimal geometry i.e. the class prototype distribution for classifying existing classes and updates it when new classes emerge and then trains a DNN model by aligning its feature space to the optimal geometry. To this end we develop a novel Dynamic Neural Collapse (DNC) algorithm to compute and update the optimal geometry. The DNC expands the geometry when new classes emerge without loss of the geometry optimality and guarantees the drift distance of old class prototypes with an explicit upper bound. Then we propose a novel Dynamic feature space Self-Organization (DYSON) method containing three major components including 1) a feature extractor 2) a Dynamic Feature-Geometry Alignment (DFGA) module aligning the feature space to the optimal geometry computed by DNC and 3) a training-free class-incremental classifier derived from the DNC geometry. Experimental comparison results on four benchmark datasets including CIFAR10 CIFAR100 CUB200 and CoRe50 demonstrate the efficiency and superiority of the DYSON method. The source code is provided in the supplementary material.

Determining the relative pose of an object between two images is pivotal to the success of generalizable object pose estimation. Existing approaches typically approximate the continuous pose representation with a large number of discrete pose hypotheses which incurs a computationally expensive process of scoring each hypothesis at test time. By contrast we present a Deep Voxel Matching Network (DVMNet) that eliminates the need for pose hypotheses and computes the relative object pose in a single pass. To this end we map the two input RGB images reference and query to their respective voxelized 3D representations. We then pass the resulting voxels through a pose estimation module where the voxels are aligned and the pose is computed in an end-to-end fashion by solving a least-squares problem. To enhance robustness we introduce a weighted closest voxel algorithm capable of mitigating the impact of noisy voxels. We conduct extensive experiments on the CO3D LINEMOD and Objaverse datasets demonstrating that our method delivers more accurate relative pose estimates for novel objects at a lower computational cost compared to state-of-the-art methods. Our code is released at: https://github.com/sailor-z/DVMNet.

Multi-view stereo reconstruction (MVS) in the wild requires to first estimate the camera intrinsic and extrinsic parameters. These are usually tedious and cumbersome to obtain yet they are mandatory to triangulate corresponding pixels in 3D space which is at the core of all best performing MVS algorithms. In this work we take an opposite stance and introduce DUSt3R a radically novel paradigm for Dense and Unconstrained Stereo 3D Reconstruction of arbitrary image collections operating without prior information about camera calibration nor viewpoint poses. We cast the pairwise reconstruction problem as a regression of pointmaps relaxing the hard constraints of usual projective camera models. We show that this formulation smoothly unifies the monocular and binocular reconstruction cases. In the case where more than two images are provided we further propose a simple yet effective global alignment strategy that expresses all pairwise pointmaps in a common reference frame. We base our network architecture on standard Transformer encoders and decoders allowing us to leverage powerful pretrained models. Our formulation directly provides a 3D model of the scene as well as depth information but interestingly we can seamlessly recover from it pixel matches focal lengths relative and absolute cameras. Extensive experiments on all these tasks showcase how DUSt3R effectively unifies various 3D vision tasks setting new performance records on monocular & multi-view depth estimation as well as relative pose estimation. In summary DUSt3R makes many geometric 3D vision tasks easy. Code and models at https://github.com/naver/dust3r

In recent years there has been a growing interest in training Neural Networks to approximate Unsigned Distance Fields (UDFs) for representing open surfaces in the context of 3D reconstruction. However UDFs are non-differentiable at the zero level set which leads to significant errors in distances and gradients generally resulting in fragmented and discontinuous surfaces. In this paper we propose to learn a hyperbolic scaling of the unsigned distance field which defines a new Eikonal problem with distinct boundary conditions. This allows our formulation to integrate seamlessly with state-of-the-art continuously differentiable implicit neural representation networks largely applied in the literature to represent signed distance fields. Our approach not only addresses the challenge of open surface representation but also demonstrates significant improvement in reconstruction quality and training performance. Moreover the unlocked field's differentiability allows the accurate computation of essential topological properties such as normal directions and curvatures pervasive in downstream tasks such as rendering. Through extensive experiments we validate our approach across various data sets and against competitive baselines. The results demonstrate enhanced accuracy and up to an order of magnitude increase in speed compared to previous methods.

Generic Face Image Quality Assessment (GFIQA) evaluates the perceptual quality of facial images which is crucial in improving image restoration algorithms and selecting high-quality face images for downstream tasks. We present a novel transformer-based method for GFIQA which is aided by two unique mechanisms. First a novel Dual-Set Degradation Representation Learning (DSL) mechanism uses facial images with both synthetic and real degradations to decouple degradation from content ensuring generalizability to real-world scenarios. This self-supervised method learns degradation features on a global scale providing a robust alternative to conventional methods that use local patch information in degradation learning. Second our transformer leverages facial landmarks to emphasize visually salient parts of a face image in evaluating its perceptual quality. We also introduce a balanced and diverse Comprehensive Generic Face IQA (CGFIQA-40k) dataset of 40K images carefully designed to overcome the biases in particular the imbalances in skin tone and gender representation in existing datasets. Extensive analysis and evaluation demonstrate the robustness of our method marking a significant improvement over prior methods.

Scene graph generation aims to capture detailed spatial and semantic relationships between objects in an image which is challenging due to incomplete labeling long-tailed relationship categories and relational semantic overlap. Existing Transformer-based methods either employ distinct queries for objects and predicates or utilize holistic queries for relation triplets and hence often suffer from limited capacity in learning low-frequency relationships. In this paper we present a new Transformer-based method called DSGG that views scene graph detection as a direct graph prediction problem based on a unique set of graph-aware queries. In particular each graph-aware query encodes a compact representation of both the node and all of its relations in the graph acquired through the utilization of a relaxed sub-graph matching during the training process. Moreover to address the problem of relational semantic overlap we utilize a strategy for relation distillation aiming to efficiently learn multiple instances of semantic relationships. Extensive experiments on the VG and the PSG datasets show that our model achieves state-of-the-art results showing a significant improvement of 3.5% and 6.7% in mR@50 and mR@100 for the scene-graph generation task and achieves an even more substantial improvement of 8.5% and 10.3% in mR@50 and mR@100 for the panoptic scene graph generation task. Code is available at https://github.com/zeeshanhayder/DSGG.

Implicit neural representations for video (NeRV) have recently become a novel way for high-quality video representation. However existing works employ a single network to represent the entire video which implicitly confuse static and dynamic information. This leads to an inability to effectively compress the redundant static information and lack the explicitly modeling of global temporal-coherent dynamic details. To solve above problems we propose DS-NeRV which decomposes videos into sparse learnable static codes and dynamic codes without the need for explicit optical flow or residual supervision. By setting different sampling rates for two codes and applying weighted sum and interpolation sampling methods DS-NeRV efficiently utilizes redundant static information while maintaining high-frequency details. Additionally we design a cross-channel attention-based (CCA) fusion module to efficiently fuse these two codes for frame decoding. Our approach achieves a high quality reconstruction of 31.2 PSNR with only 0.35M parameters thanks to separate static and dynamic codes representation and outperforms existing NeRV methods in many downstream tasks. Our project website is at https://haoyan14.github.io/DS-NeRV.

We present DRESS a large vision language model (LVLM) that innovatively exploits Natural Language feedback (NLF) from Large Language Models to enhance its alignment and interactions by addressing two key limitations in the state-of-the-art LVLMs. First prior LVLMs generally rely only on the instruction finetuning stage to enhance alignment with human preferences. Without incorporating extra feedback they are still prone to generate unhelpful hallucinated or harmful responses. Second while the visual instruction tuning data is generally structured in a multi-turn dialogue format the connections and dependencies among consecutive conversational turns are weak. This reduces the capacity for effective multi-turn interactions. To tackle these we propose a novel categorization of the NLF into two key types: critique and refinement. The critique NLF identifies the strengths and weaknesses of the responses and is used to align the LVLMs with human preferences. The refinement NLF offers concrete suggestions for improvement and is adopted to improve the interaction ability of the LVLMs-- which focuses on LVLMs' ability to refine responses by incorporating feedback in multi-turn interactions. To address the non-differentiable nature of NLF we generalize conditional reinforcement learning for training. Our experimental results demonstrate that DRESS can generate more helpful (9.76%) honest (11.52%) and harmless (21.03%) responses and more effectively learn from feedback during multi-turn interactions compared to SOTA LVLMs.

We present DREAM a novel training framework representing Diffusion Rectification and Estimation-Adaptive Models requiring minimal code changes (just three lines) yet significantly enhancing the alignment of training with sampling in diffusion models. DREAM features two components: diffusion rectification which adjusts training to reflect the sampling process and estimation adaptation which balances perception against distortion. When applied to image super-resolution (SR) DREAM adeptly navigates the tradeoff between minimizing distortion and preserving high image quality. Experiments demonstrate DREAM's superiority over standard diffusion-based SR methods showing a to faster training convergence and a to reduction in necessary sampling steps to achieve comparable or superior results. We hope DREAM will inspire a rethinking of diffusion model training paradigms.

The recovery of occluded human meshes poses challenges for current methods due to the difficulty in extracting effective image features under severe occlusion. In this paper we introduce DPMesh an innovative framework for occluded human mesh recovery that capitalizes on the profound knowledge about object structure and spatial relationships embedded in a pre-trained text-to-image diffusion model. Unlike previous methods reliant on conventional backbones for vanilla feature extraction DPMesh seamlessly integrates the pre-trained denoising U-Net with potent priors as its image backbone and performs a single-step inference to provide occlusion-aware information. To enhance the perception capability for occluded poses DPMesh incorporates judicious guidance via condition injection which produces effective controls from 2D observations for the denoising U-Net. Furthermore we explore a dedicated noisy key-point reasoning approach to mitigate disturbances arising from occlusion and crowded scenarios. This strategy fully unleashes the perceptual capability of the diffusion prior thereby enhancing accuracy. Extensive quantitative and qualitative experiments affirm the efficacy of our framework as we outperform state-of-the-art methods on both occlusion-specific and standard datasets underscoring its ability to achieve precise and robust 3D human mesh recovery particularly in challenging scenarios involving occlusion and crowded scenes. Code is available at https://github.com/EternalEvan/DPMesh.

We introduce Diffusion Parametric Head Models (DPHMs) a generative model that enables robust volumetric head reconstruction and tracking from monocular depth sequences. While recent volumetric head models such as NPHMs can now excel in representing high-fidelity head geometries tracking and reconstructing heads from real-world single-view depth sequences remains very challenging as the fitting to partial and noisy observations is underconstrained. To tackle these challenges we propose a latent diffusion-based prior to regularize volumetric head reconstruction and tracking. This prior-based regularizer effectively constrains the identity and expression codes to lie on the underlying latent manifold which represents plausible head shapes. To evaluate the effectiveness of the diffusion-based prior we collect a dataset of monocular Kinect sequences consisting of various complex facial expression motions and rapid transitions. We compare our method to state-of-the-art tracking methods and demonstrate improved head identity reconstruction as well as robust expression tracking.

Radiance fields have demonstrated impressive performance in synthesizing novel views from sparse input views yet prevailing methods suffer from high training costs and slow inference speed. This paper introduces DNGaussian a depth-regularized framework based on 3D Gaussian radiance fields offering real-time and high-quality few-shot novel view synthesis at low costs. Our motivation stems from the highly efficient representation and surprising quality of the recent 3D Gaussian Splatting despite it will encounter a geometry degradation when input views decrease. In the Gaussian radiance fields we find this degradation in scene geometry primarily lined to the positioning of Gaussian primitives and can be mitigated by depth constraint. Consequently we propose a Hard and Soft Depth Regularization to restore accurate scene geometry under coarse monocular depth supervision while maintaining a fine-grained color appearance. To further refine detailed geometry reshaping we introduce Global-Local Depth Normalization enhancing the focus on small local depth changes. Extensive experiments on LLFF DTU and Blender datasets demonstrate that DNGaussian outperforms state-of-the-art methods achieving comparable or better results with significantly reduced memory cost a 25x reduction in training time and over 3000x faster rendering speed. Code is available at: https://github.com/Fictionarry/DNGaussian

We explore visual reinforcement learning (RL) using two complementary visual modalities: frame-based RGB camera and event-based Dynamic Vision Sensor (DVS). Existing multi-modality visual RL methods often encounter challenges in effectively extracting task-relevant information from multiple modalities while suppressing the increased noise only using indirect reward signals instead of pixel-level supervision. To tackle this we propose a Decomposed Multi-Modality Representation (DMR) framework for visual RL. It explicitly decomposes the inputs into three distinct components: combined task-relevant features (co-features) RGB-specific noise and DVS-specific noise. The co-features represent the full information from both modalities that is relevant to the RL task; the two noise components each constrained by a data reconstruction loss to avoid information leak are contrasted with the co-features to maximize their difference. Extensive experiments demonstrate that by explicitly separating the different types of information our approach achieves substantially improved policy performance compared to state-of-the-art approaches.

We have witnessed significant progress in deep learning-based 3D vision ranging from neural radiance field (NeRF) based 3D representation learning to applications in novel view synthesis (NVS). However existing scene-level datasets for deep learning-based 3D vision limited to either synthetic environments or a narrow selection of real-world scenes are quite insufficient. This insufficiency not only hinders a comprehensive benchmark of existing methods but also caps what could be explored in deep learning-based 3D analysis. To address this critical gap we present DL3DV-10K a large-scale scene dataset featuring 51.2 million frames from 10510 videos captured from 65 types of point-of-interest (POI) locations covering both bounded and unbounded scenes with different levels of reflection transparency and lighting. We conducted a comprehensive benchmark of recent NVS methods on DL3DV-10K which revealed valuable insights for future research in NVS. In addition we have obtained encouraging results in a pilot study to learn generalizable NeRF from DL3DV-10K which manifests the necessity of a large-scale scene-level dataset to forge a path toward a foundation model for learning 3D representation. Our DL3DV-10K dataset benchmark results and models will be publicly accessible.

We propose a novel concept of dual and integrated latent topologies (DITTO in short) for implicit 3D reconstruction from noisy and sparse point clouds. Most existing methods predominantly focus on single latent type such as point or grid latents. In contrast the proposed DITTO leverages both point and grid latents (i.e. dual latent) to enhance their strengths the stability of grid latents and the detail-rich capability of point latents. Concretely DITTO consists of dual latent encoder and integrated implicit decoder. In the dual latent encoder a dual latent layer which is the key module block composing the encoder refines both latents in parallel maintaining their distinct shapes and enabling recursive interaction. Notably a newly proposed dynamic sparse point transformer within the dual latent layer effectively refines point latents. Then the integrated implicit decoder systematically combines these refined latents achieving high-fidelity 3D reconstruction and surpassing previous state-of-the-art methods on object- and scene-level datasets especially in thin and detailed structures.

We present DIRECT-3D a diffusion-based 3D generative model for creating high-quality 3D assets (represented by Neural Radiance Fields) from text prompts. Unlike recent 3D generative models that rely on clean and well-aligned 3D data limiting them to single or few-class generation our model is directly trained on extensive noisy and unaligned `in-the-wild' 3D assets mitigating the key challenge (i.e. data scarcity) in large-scale 3D generation. In particular DIRECT-3D is a tri-plane diffusion model that integrates two innovations: 1) A novel learning framework where noisy data are filtered and aligned automatically during the training process. Specifically after an initial warm-up phase using a small set of clean data an iterative optimization is introduced in the diffusion process to explicitly estimate the 3D pose of objects and select beneficial data based on conditional density. 2) An efficient 3D representation that is achieved by disentangling object geometry and color features with two separate conditional diffusion models that are optimized hierarchically. Given a prompt input our model generates high-quality high-resolution realistic and complex 3D objects with accurate geometric details in seconds. We achieve state-of-the-art performance in both single-class generation and text-to-3D generation. We also demonstrate that DIRECT-3D can serve as a useful 3D geometric prior of objects for example to alleviate the well-known Janus problem in 2D-lifting methods such as DreamFusion.

Instance segmentation demands substantial labeling resources. This has prompted increased interest to explore the object discovery task as an unsupervised alternative. In particular promising results were achieved in localizing instances using motion supervision only. However the motion signal introduces complexities due to its inherent noise and sparsity which constrains the effectiveness of current methodologies. In the present paper we propose DIOD (self DIstillation meets Object Discovery) the first method that places the motion-guided object discovery within a framework of continuous improvement through knowledge distillation providing solutions to existing limitations (i) DIOD robustly eliminates the noise present in the exploited motion maps providing accurate motion-supervision (ii) DIOD leverages the discovered objects within an iterative pseudo-labeling framework enriching the initial motion-supervision with static objects which results in a cost-efficient increase in performance. Through experiments on synthetic and real-world datasets we demonstrate the benefits of bridging the gap between object discovery and distillation by significantly improving the state-of-the-art. This enhancement is also sustained across other demanding metrics so far reserved for supervised tasks.

Recent advances in decentralized deep learning algorithms have demonstrated cutting-edge performance on various tasks with large pre-trained models. However a pivotal prerequisite for achieving this level of competitiveness is the significant communication and computation overheads when updating these models which prohibits the applications of them to real-world scenarios. To address this issue drawing inspiration from advanced model merging techniques without requiring additional training we introduce the Decentralized Iterative Merging-And-Training (DIMAT) paradigm--a novel decentralized deep learning framework. Within DIMAT each agent is trained on their local data and periodically merged with their neighboring agents using advanced model merging techniques like activation matching until convergence is achieved. DIMAT provably converges with the best available rate for nonconvex functions with various first-order methods while yielding tighter error bounds compared to the popular existing approaches. We conduct a comprehensive empirical analysis to validate DIMAT's superiority over baselines across diverse computer vision tasks sourced from multiple datasets. Empirical results validate our theoretical claims by showing that DIMAT attains faster and higher initial gain in accuracy with independent and identically distributed (IID) and non-IID data incurring lower communication overhead. This DIMAT paradigm presents a new opportunity for the future decentralized learning enhancing its adaptability to real-world with sparse and light-weight communication and computation.

In image question answering due to the abundant and sometimes redundant information precisely matching and integrating the information from both text and images is a challenge. In this paper we propose the Decomposition-Integration Enhancing Multimodal Insight (DIEM) which initially decomposes the given question and image into multiple subquestions and several sub-images aiming to isolate specific elements for more focused analysis. We then integrate these sub-elements by matching each subquestion with its relevant sub-images while also retaining the original image to construct a comprehensive answer to the original question without losing sight of the overall context. This strategy mirrors the human cognitive process of simplifying complex problems into smaller components for individual analysis followed by an integration of these insights. We implement DIEM on the LLaVA-v1.5 model and evaluate its performance on ScienceQA and MM-Vet. Experimental results indicate that our method boosts accuracy in most question classes of the ScienceQA (+2.03% in average) especially in the image modality (+3.40%). On MM-Vet our method achieves an improvement in MM-Vet scores increasing from 31.1 to 32.4. These findings highlight DIEM's effectiveness in harmonizing the complexities of multimodal data demonstrating its ability to enhance accuracy and depth in image question answering through its decomposition-integration process.